{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 6657,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 5.8529582023620605,
      "learning_rate": 5.0000000000000004e-08,
      "loss": 0.9245,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.512237071990967,
      "learning_rate": 1.0000000000000001e-07,
      "loss": 0.8079,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.429091930389404,
      "learning_rate": 1.5000000000000002e-07,
      "loss": 0.8043,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.227908611297607,
      "learning_rate": 2.0000000000000002e-07,
      "loss": 0.8186,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.127245903015137,
      "learning_rate": 2.5000000000000004e-07,
      "loss": 0.8398,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.284367561340332,
      "learning_rate": 3.0000000000000004e-07,
      "loss": 0.8059,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.363831520080566,
      "learning_rate": 3.5000000000000004e-07,
      "loss": 0.8135,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.5274338722229,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 0.8861,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.3875226974487305,
      "learning_rate": 4.5000000000000003e-07,
      "loss": 0.8495,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.059006690979004,
      "learning_rate": 5.000000000000001e-07,
      "loss": 0.8214,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.537599086761475,
      "learning_rate": 5.5e-07,
      "loss": 0.8752,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.7528183460235596,
      "learning_rate": 6.000000000000001e-07,
      "loss": 0.7298,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.197657585144043,
      "learning_rate": 6.5e-07,
      "loss": 0.8295,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.501471519470215,
      "learning_rate": 7.000000000000001e-07,
      "loss": 0.7799,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.451720237731934,
      "learning_rate": 7.5e-07,
      "loss": 0.847,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.214041233062744,
      "learning_rate": 8.000000000000001e-07,
      "loss": 0.8069,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.298429012298584,
      "learning_rate": 8.500000000000001e-07,
      "loss": 0.9007,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.4962332248687744,
      "learning_rate": 9.000000000000001e-07,
      "loss": 0.678,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.392894983291626,
      "learning_rate": 9.500000000000001e-07,
      "loss": 0.8114,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.3243696689605713,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.6895,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.1262621879577637,
      "learning_rate": 1.0500000000000001e-06,
      "loss": 0.7714,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.3603861331939697,
      "learning_rate": 1.1e-06,
      "loss": 0.6863,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.833707094192505,
      "learning_rate": 1.1500000000000002e-06,
      "loss": 0.7295,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.2221696376800537,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 0.5825,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.118913412094116,
      "learning_rate": 1.25e-06,
      "loss": 0.684,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.226682186126709,
      "learning_rate": 1.3e-06,
      "loss": 0.6365,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.3418819904327393,
      "learning_rate": 1.3500000000000002e-06,
      "loss": 0.5836,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1300668716430664,
      "learning_rate": 1.4000000000000001e-06,
      "loss": 0.6816,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9466854333877563,
      "learning_rate": 1.45e-06,
      "loss": 0.548,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.131221294403076,
      "learning_rate": 1.5e-06,
      "loss": 0.6301,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.537818193435669,
      "learning_rate": 1.5500000000000002e-06,
      "loss": 0.5314,
      "step": 31
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.2256839275360107,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 0.5265,
      "step": 32
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.637265205383301,
      "learning_rate": 1.6500000000000003e-06,
      "loss": 0.6232,
      "step": 33
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2843005657196045,
      "learning_rate": 1.7000000000000002e-06,
      "loss": 0.5491,
      "step": 34
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.005864381790161,
      "learning_rate": 1.75e-06,
      "loss": 0.6364,
      "step": 35
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.273056745529175,
      "learning_rate": 1.8000000000000001e-06,
      "loss": 0.5434,
      "step": 36
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.710243821144104,
      "learning_rate": 1.85e-06,
      "loss": 0.5061,
      "step": 37
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7605578899383545,
      "learning_rate": 1.9000000000000002e-06,
      "loss": 0.5577,
      "step": 38
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6441904306411743,
      "learning_rate": 1.9500000000000004e-06,
      "loss": 0.5758,
      "step": 39
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2528629302978516,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.5493,
      "step": 40
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9001293182373047,
      "learning_rate": 2.05e-06,
      "loss": 0.5489,
      "step": 41
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2381656169891357,
      "learning_rate": 2.1000000000000002e-06,
      "loss": 0.4782,
      "step": 42
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6158314943313599,
      "learning_rate": 2.15e-06,
      "loss": 0.5514,
      "step": 43
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8919001817703247,
      "learning_rate": 2.2e-06,
      "loss": 0.5748,
      "step": 44
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0388741493225098,
      "learning_rate": 2.25e-06,
      "loss": 0.5658,
      "step": 45
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.756166696548462,
      "learning_rate": 2.3000000000000004e-06,
      "loss": 0.5074,
      "step": 46
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.850195288658142,
      "learning_rate": 2.35e-06,
      "loss": 0.5131,
      "step": 47
    },
    {
      "epoch": 0.01,
      "grad_norm": 29.827350616455078,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 0.559,
      "step": 48
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7261590957641602,
      "learning_rate": 2.4500000000000003e-06,
      "loss": 0.7925,
      "step": 49
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.821045994758606,
      "learning_rate": 2.5e-06,
      "loss": 0.5486,
      "step": 50
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2954254150390625,
      "learning_rate": 2.55e-06,
      "loss": 0.4564,
      "step": 51
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6439622640609741,
      "learning_rate": 2.6e-06,
      "loss": 0.7901,
      "step": 52
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.694441080093384,
      "learning_rate": 2.6500000000000005e-06,
      "loss": 0.5378,
      "step": 53
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1778347492218018,
      "learning_rate": 2.7000000000000004e-06,
      "loss": 0.5794,
      "step": 54
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3272117376327515,
      "learning_rate": 2.7500000000000004e-06,
      "loss": 0.4907,
      "step": 55
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6430201530456543,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 0.561,
      "step": 56
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4513479471206665,
      "learning_rate": 2.85e-06,
      "loss": 0.5569,
      "step": 57
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6833441257476807,
      "learning_rate": 2.9e-06,
      "loss": 0.4999,
      "step": 58
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5854568481445312,
      "learning_rate": 2.95e-06,
      "loss": 0.4999,
      "step": 59
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5859218835830688,
      "learning_rate": 3e-06,
      "loss": 0.5311,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3431963920593262,
      "learning_rate": 3.05e-06,
      "loss": 0.4553,
      "step": 61
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7035462856292725,
      "learning_rate": 3.1000000000000004e-06,
      "loss": 0.4625,
      "step": 62
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4615570306777954,
      "learning_rate": 3.1500000000000003e-06,
      "loss": 0.4966,
      "step": 63
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.463638186454773,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 0.7819,
      "step": 64
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3215205669403076,
      "learning_rate": 3.2500000000000002e-06,
      "loss": 0.4805,
      "step": 65
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5002384185791016,
      "learning_rate": 3.3000000000000006e-06,
      "loss": 0.8236,
      "step": 66
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.561152458190918,
      "learning_rate": 3.3500000000000005e-06,
      "loss": 0.5195,
      "step": 67
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3128859996795654,
      "learning_rate": 3.4000000000000005e-06,
      "loss": 0.5174,
      "step": 68
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6727581024169922,
      "learning_rate": 3.45e-06,
      "loss": 0.5182,
      "step": 69
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6174523830413818,
      "learning_rate": 3.5e-06,
      "loss": 0.4533,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4071595668792725,
      "learning_rate": 3.5500000000000003e-06,
      "loss": 0.3998,
      "step": 71
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.515492558479309,
      "learning_rate": 3.6000000000000003e-06,
      "loss": 0.5114,
      "step": 72
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.217383861541748,
      "learning_rate": 3.65e-06,
      "loss": 0.737,
      "step": 73
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.030747890472412,
      "learning_rate": 3.7e-06,
      "loss": 0.4751,
      "step": 74
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6317182779312134,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 0.5379,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5583616495132446,
      "learning_rate": 3.8000000000000005e-06,
      "loss": 0.4856,
      "step": 76
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6841261386871338,
      "learning_rate": 3.85e-06,
      "loss": 0.5736,
      "step": 77
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.445110559463501,
      "learning_rate": 3.900000000000001e-06,
      "loss": 0.5113,
      "step": 78
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.167898654937744,
      "learning_rate": 3.95e-06,
      "loss": 0.5349,
      "step": 79
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.422402024269104,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.5364,
      "step": 80
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6843194961547852,
      "learning_rate": 4.05e-06,
      "loss": 0.4743,
      "step": 81
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4669547080993652,
      "learning_rate": 4.1e-06,
      "loss": 0.6087,
      "step": 82
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5706548690795898,
      "learning_rate": 4.15e-06,
      "loss": 0.5885,
      "step": 83
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.329118251800537,
      "learning_rate": 4.2000000000000004e-06,
      "loss": 0.4728,
      "step": 84
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5949207544326782,
      "learning_rate": 4.25e-06,
      "loss": 0.4611,
      "step": 85
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5131396055221558,
      "learning_rate": 4.3e-06,
      "loss": 0.4858,
      "step": 86
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9135791063308716,
      "learning_rate": 4.350000000000001e-06,
      "loss": 0.6025,
      "step": 87
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5825953483581543,
      "learning_rate": 4.4e-06,
      "loss": 0.4651,
      "step": 88
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6479098796844482,
      "learning_rate": 4.450000000000001e-06,
      "loss": 0.4298,
      "step": 89
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.792706847190857,
      "learning_rate": 4.5e-06,
      "loss": 0.4459,
      "step": 90
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.341231346130371,
      "learning_rate": 4.5500000000000005e-06,
      "loss": 0.5218,
      "step": 91
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5417590141296387,
      "learning_rate": 4.600000000000001e-06,
      "loss": 0.4646,
      "step": 92
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4662986993789673,
      "learning_rate": 4.65e-06,
      "loss": 0.5729,
      "step": 93
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.138056993484497,
      "learning_rate": 4.7e-06,
      "loss": 0.4938,
      "step": 94
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5889183282852173,
      "learning_rate": 4.75e-06,
      "loss": 0.4181,
      "step": 95
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5066238641738892,
      "learning_rate": 4.800000000000001e-06,
      "loss": 0.7882,
      "step": 96
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5680415630340576,
      "learning_rate": 4.85e-06,
      "loss": 0.4851,
      "step": 97
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.569085955619812,
      "learning_rate": 4.9000000000000005e-06,
      "loss": 0.5427,
      "step": 98
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0821168422698975,
      "learning_rate": 4.95e-06,
      "loss": 0.5405,
      "step": 99
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8579951524734497,
      "learning_rate": 5e-06,
      "loss": 0.5092,
      "step": 100
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7069708108901978,
      "learning_rate": 5.050000000000001e-06,
      "loss": 0.4556,
      "step": 101
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4003320932388306,
      "learning_rate": 5.1e-06,
      "loss": 0.8093,
      "step": 102
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4428329467773438,
      "learning_rate": 5.150000000000001e-06,
      "loss": 0.5274,
      "step": 103
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1690846681594849,
      "learning_rate": 5.2e-06,
      "loss": 0.728,
      "step": 104
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6877247095108032,
      "learning_rate": 5.2500000000000006e-06,
      "loss": 0.4241,
      "step": 105
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1933305263519287,
      "learning_rate": 5.300000000000001e-06,
      "loss": 0.7562,
      "step": 106
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6397169828414917,
      "learning_rate": 5.3500000000000004e-06,
      "loss": 0.4716,
      "step": 107
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7164292335510254,
      "learning_rate": 5.400000000000001e-06,
      "loss": 0.4798,
      "step": 108
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5259934663772583,
      "learning_rate": 5.450000000000001e-06,
      "loss": 0.5005,
      "step": 109
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.201320171356201,
      "learning_rate": 5.500000000000001e-06,
      "loss": 0.4551,
      "step": 110
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5878087282180786,
      "learning_rate": 5.550000000000001e-06,
      "loss": 0.4687,
      "step": 111
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.237149715423584,
      "learning_rate": 5.600000000000001e-06,
      "loss": 0.5181,
      "step": 112
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3826124668121338,
      "learning_rate": 5.65e-06,
      "loss": 0.4749,
      "step": 113
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9310431480407715,
      "learning_rate": 5.7e-06,
      "loss": 0.5002,
      "step": 114
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4320611953735352,
      "learning_rate": 5.75e-06,
      "loss": 0.5203,
      "step": 115
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3641215562820435,
      "learning_rate": 5.8e-06,
      "loss": 0.4728,
      "step": 116
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7420451641082764,
      "learning_rate": 5.85e-06,
      "loss": 0.51,
      "step": 117
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9351675510406494,
      "learning_rate": 5.9e-06,
      "loss": 0.5415,
      "step": 118
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8387118577957153,
      "learning_rate": 5.950000000000001e-06,
      "loss": 0.4723,
      "step": 119
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6368470191955566,
      "learning_rate": 6e-06,
      "loss": 0.4916,
      "step": 120
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7157363891601562,
      "learning_rate": 6.0500000000000005e-06,
      "loss": 0.534,
      "step": 121
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4537992477416992,
      "learning_rate": 6.1e-06,
      "loss": 0.7489,
      "step": 122
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3560537099838257,
      "learning_rate": 6.15e-06,
      "loss": 0.4299,
      "step": 123
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.057404041290283,
      "learning_rate": 6.200000000000001e-06,
      "loss": 0.4394,
      "step": 124
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.423816442489624,
      "learning_rate": 6.25e-06,
      "loss": 0.4852,
      "step": 125
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3739547729492188,
      "learning_rate": 6.300000000000001e-06,
      "loss": 0.5268,
      "step": 126
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6192394495010376,
      "learning_rate": 6.35e-06,
      "loss": 0.4433,
      "step": 127
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.734795093536377,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 0.4485,
      "step": 128
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2830787897109985,
      "learning_rate": 6.450000000000001e-06,
      "loss": 0.5244,
      "step": 129
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4225298166275024,
      "learning_rate": 6.5000000000000004e-06,
      "loss": 0.7207,
      "step": 130
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8243030309677124,
      "learning_rate": 6.550000000000001e-06,
      "loss": 0.4688,
      "step": 131
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4373445510864258,
      "learning_rate": 6.600000000000001e-06,
      "loss": 0.5019,
      "step": 132
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.053093433380127,
      "learning_rate": 6.650000000000001e-06,
      "loss": 0.4993,
      "step": 133
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6475763320922852,
      "learning_rate": 6.700000000000001e-06,
      "loss": 0.5152,
      "step": 134
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.412107229232788,
      "learning_rate": 6.750000000000001e-06,
      "loss": 0.7628,
      "step": 135
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9530085325241089,
      "learning_rate": 6.800000000000001e-06,
      "loss": 0.525,
      "step": 136
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.806856632232666,
      "learning_rate": 6.850000000000001e-06,
      "loss": 0.4165,
      "step": 137
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.463667392730713,
      "learning_rate": 6.9e-06,
      "loss": 0.5152,
      "step": 138
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4566363096237183,
      "learning_rate": 6.95e-06,
      "loss": 0.4832,
      "step": 139
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.751391887664795,
      "learning_rate": 7e-06,
      "loss": 0.5394,
      "step": 140
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.636208176612854,
      "learning_rate": 7.05e-06,
      "loss": 0.493,
      "step": 141
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8021098375320435,
      "learning_rate": 7.100000000000001e-06,
      "loss": 0.5002,
      "step": 142
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3746438026428223,
      "learning_rate": 7.15e-06,
      "loss": 0.4516,
      "step": 143
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.9152238368988037,
      "learning_rate": 7.2000000000000005e-06,
      "loss": 0.603,
      "step": 144
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1137917041778564,
      "learning_rate": 7.25e-06,
      "loss": 0.4676,
      "step": 145
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.444410800933838,
      "learning_rate": 7.3e-06,
      "loss": 0.4542,
      "step": 146
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.780739665031433,
      "learning_rate": 7.350000000000001e-06,
      "loss": 0.4397,
      "step": 147
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7790647745132446,
      "learning_rate": 7.4e-06,
      "loss": 0.5106,
      "step": 148
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9565337896347046,
      "learning_rate": 7.450000000000001e-06,
      "loss": 0.453,
      "step": 149
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.746363878250122,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.4641,
      "step": 150
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4024478197097778,
      "learning_rate": 7.5500000000000006e-06,
      "loss": 0.5385,
      "step": 151
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.32373332977295,
      "learning_rate": 7.600000000000001e-06,
      "loss": 0.4924,
      "step": 152
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6879143714904785,
      "learning_rate": 7.650000000000001e-06,
      "loss": 0.4463,
      "step": 153
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7326459884643555,
      "learning_rate": 7.7e-06,
      "loss": 0.4862,
      "step": 154
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3022948503494263,
      "learning_rate": 7.75e-06,
      "loss": 0.4469,
      "step": 155
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.683469533920288,
      "learning_rate": 7.800000000000002e-06,
      "loss": 0.4561,
      "step": 156
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6714413166046143,
      "learning_rate": 7.850000000000001e-06,
      "loss": 0.4525,
      "step": 157
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.630603313446045,
      "learning_rate": 7.9e-06,
      "loss": 0.4572,
      "step": 158
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7316653728485107,
      "learning_rate": 7.950000000000002e-06,
      "loss": 0.4623,
      "step": 159
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.597062349319458,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.492,
      "step": 160
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7099560499191284,
      "learning_rate": 8.050000000000001e-06,
      "loss": 0.453,
      "step": 161
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6483113765716553,
      "learning_rate": 8.1e-06,
      "loss": 0.4265,
      "step": 162
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3760236501693726,
      "learning_rate": 8.15e-06,
      "loss": 0.4679,
      "step": 163
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9463396072387695,
      "learning_rate": 8.2e-06,
      "loss": 0.4928,
      "step": 164
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5747789144515991,
      "learning_rate": 8.25e-06,
      "loss": 0.563,
      "step": 165
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5490456819534302,
      "learning_rate": 8.3e-06,
      "loss": 0.5067,
      "step": 166
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6436101198196411,
      "learning_rate": 8.35e-06,
      "loss": 0.4996,
      "step": 167
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2492573261260986,
      "learning_rate": 8.400000000000001e-06,
      "loss": 0.429,
      "step": 168
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.348167061805725,
      "learning_rate": 8.45e-06,
      "loss": 0.5024,
      "step": 169
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.741473913192749,
      "learning_rate": 8.5e-06,
      "loss": 0.5102,
      "step": 170
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.934858322143555,
      "learning_rate": 8.550000000000001e-06,
      "loss": 0.4682,
      "step": 171
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1638978719711304,
      "learning_rate": 8.6e-06,
      "loss": 0.4452,
      "step": 172
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6449917554855347,
      "learning_rate": 8.65e-06,
      "loss": 0.3827,
      "step": 173
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5274101495742798,
      "learning_rate": 8.700000000000001e-06,
      "loss": 0.534,
      "step": 174
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5389479398727417,
      "learning_rate": 8.750000000000001e-06,
      "loss": 0.5034,
      "step": 175
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1131751537323,
      "learning_rate": 8.8e-06,
      "loss": 0.5151,
      "step": 176
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9604549407958984,
      "learning_rate": 8.85e-06,
      "loss": 0.5087,
      "step": 177
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7045478820800781,
      "learning_rate": 8.900000000000001e-06,
      "loss": 0.4996,
      "step": 178
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4331666231155396,
      "learning_rate": 8.95e-06,
      "loss": 0.4747,
      "step": 179
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4885194301605225,
      "learning_rate": 9e-06,
      "loss": 0.4366,
      "step": 180
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4685351848602295,
      "learning_rate": 9.050000000000001e-06,
      "loss": 0.4378,
      "step": 181
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.9353721141815186,
      "learning_rate": 9.100000000000001e-06,
      "loss": 0.4874,
      "step": 182
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4415245056152344,
      "learning_rate": 9.15e-06,
      "loss": 0.4646,
      "step": 183
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.245835542678833,
      "learning_rate": 9.200000000000002e-06,
      "loss": 0.4803,
      "step": 184
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.253498911857605,
      "learning_rate": 9.250000000000001e-06,
      "loss": 0.5269,
      "step": 185
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6602411270141602,
      "learning_rate": 9.3e-06,
      "loss": 0.5429,
      "step": 186
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.284512996673584,
      "learning_rate": 9.350000000000002e-06,
      "loss": 0.5649,
      "step": 187
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4753460884094238,
      "learning_rate": 9.4e-06,
      "loss": 0.4165,
      "step": 188
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.456650972366333,
      "learning_rate": 9.450000000000001e-06,
      "loss": 0.4885,
      "step": 189
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.283982276916504,
      "learning_rate": 9.5e-06,
      "loss": 0.4884,
      "step": 190
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.688179612159729,
      "learning_rate": 9.55e-06,
      "loss": 0.5298,
      "step": 191
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3304102420806885,
      "learning_rate": 9.600000000000001e-06,
      "loss": 0.4312,
      "step": 192
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8922255039215088,
      "learning_rate": 9.65e-06,
      "loss": 0.4804,
      "step": 193
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2636334896087646,
      "learning_rate": 9.7e-06,
      "loss": 0.4486,
      "step": 194
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9050902128219604,
      "learning_rate": 9.75e-06,
      "loss": 0.501,
      "step": 195
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6458041667938232,
      "learning_rate": 9.800000000000001e-06,
      "loss": 0.4677,
      "step": 196
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.676837921142578,
      "learning_rate": 9.85e-06,
      "loss": 0.5956,
      "step": 197
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3418490886688232,
      "learning_rate": 9.9e-06,
      "loss": 0.487,
      "step": 198
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.639165997505188,
      "learning_rate": 9.950000000000001e-06,
      "loss": 0.4731,
      "step": 199
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3942128419876099,
      "learning_rate": 1e-05,
      "loss": 0.469,
      "step": 200
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4801722764968872,
      "learning_rate": 9.999999408195624e-06,
      "loss": 0.7471,
      "step": 201
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2643400430679321,
      "learning_rate": 9.999997632782634e-06,
      "loss": 0.4914,
      "step": 202
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5594366788864136,
      "learning_rate": 9.999994673761451e-06,
      "loss": 0.4845,
      "step": 203
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.827388048171997,
      "learning_rate": 9.999990531132775e-06,
      "loss": 0.4401,
      "step": 204
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0800230503082275,
      "learning_rate": 9.999985204897588e-06,
      "loss": 0.4408,
      "step": 205
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5237919092178345,
      "learning_rate": 9.999978695057148e-06,
      "loss": 0.4801,
      "step": 206
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5008735656738281,
      "learning_rate": 9.999971001613e-06,
      "loss": 0.5318,
      "step": 207
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7397962808609009,
      "learning_rate": 9.999962124566963e-06,
      "loss": 0.4718,
      "step": 208
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.491859793663025,
      "learning_rate": 9.999952063921136e-06,
      "loss": 0.4443,
      "step": 209
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.295931816101074,
      "learning_rate": 9.999940819677906e-06,
      "loss": 0.5291,
      "step": 210
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3922126293182373,
      "learning_rate": 9.99992839183993e-06,
      "loss": 0.4901,
      "step": 211
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4824954271316528,
      "learning_rate": 9.999914780410152e-06,
      "loss": 0.5228,
      "step": 212
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3611721992492676,
      "learning_rate": 9.999899985391796e-06,
      "loss": 0.772,
      "step": 213
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.341275930404663,
      "learning_rate": 9.99988400678836e-06,
      "loss": 0.5052,
      "step": 214
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5510364770889282,
      "learning_rate": 9.999866844603629e-06,
      "loss": 0.5493,
      "step": 215
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.417754888534546,
      "learning_rate": 9.999848498841665e-06,
      "loss": 0.5212,
      "step": 216
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4278830289840698,
      "learning_rate": 9.999828969506814e-06,
      "loss": 0.4777,
      "step": 217
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.384564757347107,
      "learning_rate": 9.999808256603692e-06,
      "loss": 0.5084,
      "step": 218
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2734190225601196,
      "learning_rate": 9.999786360137208e-06,
      "loss": 0.5223,
      "step": 219
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.158571243286133,
      "learning_rate": 9.999763280112545e-06,
      "loss": 0.4724,
      "step": 220
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.061119794845581,
      "learning_rate": 9.999739016535164e-06,
      "loss": 0.5062,
      "step": 221
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0797109603881836,
      "learning_rate": 9.99971356941081e-06,
      "loss": 0.5126,
      "step": 222
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3694355487823486,
      "learning_rate": 9.999686938745508e-06,
      "loss": 0.7898,
      "step": 223
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6570559740066528,
      "learning_rate": 9.99965912454556e-06,
      "loss": 0.4916,
      "step": 224
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1357316970825195,
      "learning_rate": 9.999630126817553e-06,
      "loss": 0.4461,
      "step": 225
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2560936212539673,
      "learning_rate": 9.999599945568348e-06,
      "loss": 0.4961,
      "step": 226
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1632850170135498,
      "learning_rate": 9.99956858080509e-06,
      "loss": 0.4959,
      "step": 227
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2387595176696777,
      "learning_rate": 9.999536032535206e-06,
      "loss": 0.5258,
      "step": 228
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.789397954940796,
      "learning_rate": 9.9995023007664e-06,
      "loss": 0.5036,
      "step": 229
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.638387680053711,
      "learning_rate": 9.999467385506659e-06,
      "loss": 0.5394,
      "step": 230
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3916290998458862,
      "learning_rate": 9.999431286764244e-06,
      "loss": 0.5668,
      "step": 231
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3547253608703613,
      "learning_rate": 9.9993940045477e-06,
      "loss": 0.4806,
      "step": 232
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7931426763534546,
      "learning_rate": 9.999355538865859e-06,
      "loss": 0.4485,
      "step": 233
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4922987222671509,
      "learning_rate": 9.99931588972782e-06,
      "loss": 0.4662,
      "step": 234
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3328020572662354,
      "learning_rate": 9.999275057142971e-06,
      "loss": 0.4289,
      "step": 235
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9149820804595947,
      "learning_rate": 9.999233041120979e-06,
      "loss": 0.4835,
      "step": 236
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.395357131958008,
      "learning_rate": 9.99918984167179e-06,
      "loss": 0.5021,
      "step": 237
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3180311918258667,
      "learning_rate": 9.999145458805628e-06,
      "loss": 0.4192,
      "step": 238
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.105564117431641,
      "learning_rate": 9.999099892533004e-06,
      "loss": 0.4422,
      "step": 239
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6254515647888184,
      "learning_rate": 9.999053142864699e-06,
      "loss": 0.5166,
      "step": 240
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2186214923858643,
      "learning_rate": 9.999005209811783e-06,
      "loss": 0.3995,
      "step": 241
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2826241254806519,
      "learning_rate": 9.998956093385603e-06,
      "loss": 0.4175,
      "step": 242
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.340816855430603,
      "learning_rate": 9.998905793597784e-06,
      "loss": 0.5016,
      "step": 243
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6064252853393555,
      "learning_rate": 9.998854310460233e-06,
      "loss": 0.5284,
      "step": 244
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.8882603645324707,
      "learning_rate": 9.998801643985139e-06,
      "loss": 0.4722,
      "step": 245
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6175823211669922,
      "learning_rate": 9.99874779418497e-06,
      "loss": 0.4947,
      "step": 246
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5915039777755737,
      "learning_rate": 9.99869276107247e-06,
      "loss": 0.4363,
      "step": 247
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.25153386592865,
      "learning_rate": 9.998636544660669e-06,
      "loss": 0.4899,
      "step": 248
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2472151517868042,
      "learning_rate": 9.998579144962874e-06,
      "loss": 0.5004,
      "step": 249
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6414834260940552,
      "learning_rate": 9.998520561992675e-06,
      "loss": 0.4867,
      "step": 250
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5103343725204468,
      "learning_rate": 9.998460795763935e-06,
      "loss": 0.5135,
      "step": 251
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3180837631225586,
      "learning_rate": 9.998399846290805e-06,
      "loss": 0.4916,
      "step": 252
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3299529552459717,
      "learning_rate": 9.998337713587714e-06,
      "loss": 0.4816,
      "step": 253
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.68578040599823,
      "learning_rate": 9.99827439766937e-06,
      "loss": 0.448,
      "step": 254
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6348130702972412,
      "learning_rate": 9.998209898550756e-06,
      "loss": 0.575,
      "step": 255
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.394629955291748,
      "learning_rate": 9.998144216247148e-06,
      "loss": 0.4442,
      "step": 256
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8812670707702637,
      "learning_rate": 9.99807735077409e-06,
      "loss": 0.491,
      "step": 257
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3125371932983398,
      "learning_rate": 9.99800930214741e-06,
      "loss": 0.4759,
      "step": 258
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.593997597694397,
      "learning_rate": 9.997940070383221e-06,
      "loss": 0.498,
      "step": 259
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7125550508499146,
      "learning_rate": 9.997869655497906e-06,
      "loss": 0.4867,
      "step": 260
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7238202095031738,
      "learning_rate": 9.997798057508139e-06,
      "loss": 0.4273,
      "step": 261
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1468383073806763,
      "learning_rate": 9.997725276430863e-06,
      "loss": 0.7332,
      "step": 262
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.322562336921692,
      "learning_rate": 9.997651312283313e-06,
      "loss": 0.4874,
      "step": 263
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3481788635253906,
      "learning_rate": 9.997576165082994e-06,
      "loss": 0.4451,
      "step": 264
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6522122621536255,
      "learning_rate": 9.997499834847697e-06,
      "loss": 0.4988,
      "step": 265
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0371108055114746,
      "learning_rate": 9.997422321595488e-06,
      "loss": 0.424,
      "step": 266
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6888210773468018,
      "learning_rate": 9.99734362534472e-06,
      "loss": 0.4769,
      "step": 267
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1439499855041504,
      "learning_rate": 9.997263746114019e-06,
      "loss": 0.4803,
      "step": 268
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.555004358291626,
      "learning_rate": 9.997182683922295e-06,
      "loss": 0.5753,
      "step": 269
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5724085569381714,
      "learning_rate": 9.997100438788738e-06,
      "loss": 0.5017,
      "step": 270
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3484128713607788,
      "learning_rate": 9.997017010732818e-06,
      "loss": 0.5805,
      "step": 271
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7690327167510986,
      "learning_rate": 9.996932399774282e-06,
      "loss": 0.4908,
      "step": 272
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.298746585845947,
      "learning_rate": 9.99684660593316e-06,
      "loss": 0.4965,
      "step": 273
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.550227403640747,
      "learning_rate": 9.996759629229762e-06,
      "loss": 0.431,
      "step": 274
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0221312046051025,
      "learning_rate": 9.996671469684676e-06,
      "loss": 0.5,
      "step": 275
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.344545602798462,
      "learning_rate": 9.996582127318773e-06,
      "loss": 0.5717,
      "step": 276
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2620075941085815,
      "learning_rate": 9.996491602153202e-06,
      "loss": 0.4615,
      "step": 277
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9680190086364746,
      "learning_rate": 9.99639989420939e-06,
      "loss": 0.4511,
      "step": 278
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.563599944114685,
      "learning_rate": 9.99630700350905e-06,
      "loss": 0.4968,
      "step": 279
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2936493158340454,
      "learning_rate": 9.996212930074167e-06,
      "loss": 0.7424,
      "step": 280
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.325316309928894,
      "learning_rate": 9.996117673927013e-06,
      "loss": 0.4876,
      "step": 281
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.461725115776062,
      "learning_rate": 9.996021235090137e-06,
      "loss": 0.4576,
      "step": 282
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4117025136947632,
      "learning_rate": 9.995923613586367e-06,
      "loss": 0.81,
      "step": 283
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.254095196723938,
      "learning_rate": 9.995824809438814e-06,
      "loss": 0.5181,
      "step": 284
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3672224283218384,
      "learning_rate": 9.995724822670866e-06,
      "loss": 0.5091,
      "step": 285
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.78434419631958,
      "learning_rate": 9.995623653306191e-06,
      "loss": 0.5061,
      "step": 286
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2910066843032837,
      "learning_rate": 9.995521301368741e-06,
      "loss": 0.4952,
      "step": 287
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3810553550720215,
      "learning_rate": 9.99541776688274e-06,
      "loss": 0.4542,
      "step": 288
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5316358804702759,
      "learning_rate": 9.995313049872701e-06,
      "loss": 0.4958,
      "step": 289
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.359970211982727,
      "learning_rate": 9.995207150363413e-06,
      "loss": 0.4908,
      "step": 290
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3316508531570435,
      "learning_rate": 9.995100068379942e-06,
      "loss": 0.5217,
      "step": 291
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.6324825286865234,
      "learning_rate": 9.994991803947638e-06,
      "loss": 0.5121,
      "step": 292
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.452622413635254,
      "learning_rate": 9.99488235709213e-06,
      "loss": 0.4833,
      "step": 293
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.109365463256836,
      "learning_rate": 9.994771727839325e-06,
      "loss": 0.4096,
      "step": 294
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5960341691970825,
      "learning_rate": 9.994659916215414e-06,
      "loss": 0.476,
      "step": 295
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.4579899311065674,
      "learning_rate": 9.994546922246862e-06,
      "loss": 0.4505,
      "step": 296
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6756395101547241,
      "learning_rate": 9.994432745960419e-06,
      "loss": 0.5175,
      "step": 297
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4486167430877686,
      "learning_rate": 9.994317387383112e-06,
      "loss": 0.5264,
      "step": 298
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6505334377288818,
      "learning_rate": 9.994200846542251e-06,
      "loss": 0.5713,
      "step": 299
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2904553413391113,
      "learning_rate": 9.994083123465423e-06,
      "loss": 0.4558,
      "step": 300
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2456185817718506,
      "learning_rate": 9.993964218180493e-06,
      "loss": 0.4804,
      "step": 301
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6007000207901,
      "learning_rate": 9.993844130715611e-06,
      "loss": 0.5786,
      "step": 302
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.378096342086792,
      "learning_rate": 9.993722861099204e-06,
      "loss": 0.4061,
      "step": 303
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3088324069976807,
      "learning_rate": 9.993600409359978e-06,
      "loss": 0.4517,
      "step": 304
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.268105387687683,
      "learning_rate": 9.993476775526922e-06,
      "loss": 0.4783,
      "step": 305
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.934381365776062,
      "learning_rate": 9.993351959629302e-06,
      "loss": 0.46,
      "step": 306
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6831387281417847,
      "learning_rate": 9.993225961696663e-06,
      "loss": 0.4853,
      "step": 307
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1404776573181152,
      "learning_rate": 9.993098781758834e-06,
      "loss": 0.5048,
      "step": 308
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2501282691955566,
      "learning_rate": 9.99297041984592e-06,
      "loss": 0.5459,
      "step": 309
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3752021789550781,
      "learning_rate": 9.992840875988306e-06,
      "loss": 0.7559,
      "step": 310
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3770265579223633,
      "learning_rate": 9.992710150216657e-06,
      "loss": 0.4655,
      "step": 311
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4042694568634033,
      "learning_rate": 9.992578242561925e-06,
      "loss": 0.5169,
      "step": 312
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5329980850219727,
      "learning_rate": 9.992445153055327e-06,
      "loss": 0.4716,
      "step": 313
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1964659690856934,
      "learning_rate": 9.992310881728375e-06,
      "loss": 0.457,
      "step": 314
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5566719770431519,
      "learning_rate": 9.992175428612851e-06,
      "loss": 0.5215,
      "step": 315
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.991162657737732,
      "learning_rate": 9.99203879374082e-06,
      "loss": 0.5057,
      "step": 316
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.714535117149353,
      "learning_rate": 9.991900977144625e-06,
      "loss": 0.5278,
      "step": 317
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6516565084457397,
      "learning_rate": 9.991761978856891e-06,
      "loss": 0.5448,
      "step": 318
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.217052698135376,
      "learning_rate": 9.991621798910522e-06,
      "loss": 0.5,
      "step": 319
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1665018796920776,
      "learning_rate": 9.991480437338704e-06,
      "loss": 0.7059,
      "step": 320
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.42246675491333,
      "learning_rate": 9.991337894174898e-06,
      "loss": 0.5159,
      "step": 321
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4899076223373413,
      "learning_rate": 9.991194169452846e-06,
      "loss": 0.4444,
      "step": 322
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.490022897720337,
      "learning_rate": 9.991049263206573e-06,
      "loss": 0.7065,
      "step": 323
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2754420042037964,
      "learning_rate": 9.99090317547038e-06,
      "loss": 0.4215,
      "step": 324
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9068132638931274,
      "learning_rate": 9.990755906278851e-06,
      "loss": 0.516,
      "step": 325
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.754051446914673,
      "learning_rate": 9.990607455666847e-06,
      "loss": 0.5488,
      "step": 326
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.376944899559021,
      "learning_rate": 9.990457823669508e-06,
      "loss": 0.5524,
      "step": 327
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5095914602279663,
      "learning_rate": 9.990307010322256e-06,
      "loss": 0.5301,
      "step": 328
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3423576354980469,
      "learning_rate": 9.990155015660794e-06,
      "loss": 0.4896,
      "step": 329
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5664844512939453,
      "learning_rate": 9.990001839721097e-06,
      "loss": 0.4487,
      "step": 330
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.361722469329834,
      "learning_rate": 9.989847482539433e-06,
      "loss": 0.5036,
      "step": 331
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.56172513961792,
      "learning_rate": 9.989691944152333e-06,
      "loss": 0.521,
      "step": 332
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7463650703430176,
      "learning_rate": 9.989535224596623e-06,
      "loss": 0.4914,
      "step": 333
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1691831350326538,
      "learning_rate": 9.989377323909399e-06,
      "loss": 0.7169,
      "step": 334
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3612710237503052,
      "learning_rate": 9.98921824212804e-06,
      "loss": 0.4734,
      "step": 335
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2443385124206543,
      "learning_rate": 9.989057979290204e-06,
      "loss": 0.7189,
      "step": 336
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4218955039978027,
      "learning_rate": 9.98889653543383e-06,
      "loss": 0.507,
      "step": 337
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2722761631011963,
      "learning_rate": 9.988733910597132e-06,
      "loss": 0.4493,
      "step": 338
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.934725046157837,
      "learning_rate": 9.98857010481861e-06,
      "loss": 0.4035,
      "step": 339
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3017216920852661,
      "learning_rate": 9.98840511813704e-06,
      "loss": 0.4406,
      "step": 340
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8587690591812134,
      "learning_rate": 9.988238950591478e-06,
      "loss": 0.5523,
      "step": 341
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.893813967704773,
      "learning_rate": 9.988071602221256e-06,
      "loss": 0.459,
      "step": 342
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.320626974105835,
      "learning_rate": 9.987903073065995e-06,
      "loss": 0.5279,
      "step": 343
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8401105403900146,
      "learning_rate": 9.987733363165584e-06,
      "loss": 0.4366,
      "step": 344
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9724891185760498,
      "learning_rate": 9.9875624725602e-06,
      "loss": 0.5576,
      "step": 345
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.046355962753296,
      "learning_rate": 9.987390401290295e-06,
      "loss": 0.587,
      "step": 346
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.571437954902649,
      "learning_rate": 9.987217149396605e-06,
      "loss": 0.5012,
      "step": 347
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4487056732177734,
      "learning_rate": 9.987042716920138e-06,
      "loss": 0.5209,
      "step": 348
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.52364981174469,
      "learning_rate": 9.98686710390219e-06,
      "loss": 0.4645,
      "step": 349
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5838154554367065,
      "learning_rate": 9.986690310384329e-06,
      "loss": 0.4712,
      "step": 350
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8044450283050537,
      "learning_rate": 9.98651233640841e-06,
      "loss": 0.489,
      "step": 351
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.845239520072937,
      "learning_rate": 9.986333182016559e-06,
      "loss": 0.4758,
      "step": 352
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5945872068405151,
      "learning_rate": 9.986152847251187e-06,
      "loss": 0.4703,
      "step": 353
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3067760467529297,
      "learning_rate": 9.985971332154985e-06,
      "loss": 0.7633,
      "step": 354
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3014905452728271,
      "learning_rate": 9.98578863677092e-06,
      "loss": 0.6925,
      "step": 355
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2983912229537964,
      "learning_rate": 9.985604761142242e-06,
      "loss": 0.4811,
      "step": 356
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.514220714569092,
      "learning_rate": 9.985419705312475e-06,
      "loss": 0.4248,
      "step": 357
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4145488739013672,
      "learning_rate": 9.985233469325429e-06,
      "loss": 0.4,
      "step": 358
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8904967308044434,
      "learning_rate": 9.985046053225186e-06,
      "loss": 0.5147,
      "step": 359
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8366272449493408,
      "learning_rate": 9.984857457056116e-06,
      "loss": 0.5016,
      "step": 360
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4698734283447266,
      "learning_rate": 9.984667680862861e-06,
      "loss": 0.5708,
      "step": 361
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1643868684768677,
      "learning_rate": 9.984476724690346e-06,
      "loss": 0.3737,
      "step": 362
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2029333114624023,
      "learning_rate": 9.984284588583774e-06,
      "loss": 0.5243,
      "step": 363
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1546447277069092,
      "learning_rate": 9.98409127258863e-06,
      "loss": 0.4348,
      "step": 364
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.202029228210449,
      "learning_rate": 9.983896776750673e-06,
      "loss": 0.5462,
      "step": 365
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3566277027130127,
      "learning_rate": 9.983701101115946e-06,
      "loss": 0.5463,
      "step": 366
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4920955896377563,
      "learning_rate": 9.983504245730771e-06,
      "loss": 0.5777,
      "step": 367
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.307928204536438,
      "learning_rate": 9.983306210641744e-06,
      "loss": 0.5739,
      "step": 368
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2147247791290283,
      "learning_rate": 9.983106995895748e-06,
      "loss": 0.4259,
      "step": 369
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3956869840621948,
      "learning_rate": 9.98290660153994e-06,
      "loss": 0.4841,
      "step": 370
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.360257625579834,
      "learning_rate": 9.982705027621757e-06,
      "loss": 0.494,
      "step": 371
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3886780738830566,
      "learning_rate": 9.982502274188917e-06,
      "loss": 0.5525,
      "step": 372
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3264962434768677,
      "learning_rate": 9.982298341289416e-06,
      "loss": 0.4703,
      "step": 373
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4636390209197998,
      "learning_rate": 9.98209322897153e-06,
      "loss": 0.4464,
      "step": 374
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.093502163887024,
      "learning_rate": 9.981886937283812e-06,
      "loss": 0.4749,
      "step": 375
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3921705484390259,
      "learning_rate": 9.981679466275095e-06,
      "loss": 0.4778,
      "step": 376
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.379813551902771,
      "learning_rate": 9.981470815994496e-06,
      "loss": 0.4665,
      "step": 377
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4467246532440186,
      "learning_rate": 9.981260986491404e-06,
      "loss": 0.745,
      "step": 378
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6330857276916504,
      "learning_rate": 9.981049977815489e-06,
      "loss": 0.5276,
      "step": 379
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8342814445495605,
      "learning_rate": 9.980837790016704e-06,
      "loss": 0.4934,
      "step": 380
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1055431365966797,
      "learning_rate": 9.980624423145278e-06,
      "loss": 0.5607,
      "step": 381
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.191233515739441,
      "learning_rate": 9.980409877251718e-06,
      "loss": 0.4617,
      "step": 382
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4384244680404663,
      "learning_rate": 9.980194152386813e-06,
      "loss": 0.4639,
      "step": 383
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9799892902374268,
      "learning_rate": 9.979977248601629e-06,
      "loss": 0.3943,
      "step": 384
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1561627388000488,
      "learning_rate": 9.979759165947514e-06,
      "loss": 0.4615,
      "step": 385
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7912955284118652,
      "learning_rate": 9.979539904476089e-06,
      "loss": 0.5758,
      "step": 386
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.464391827583313,
      "learning_rate": 9.979319464239263e-06,
      "loss": 0.4269,
      "step": 387
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.2505042552948,
      "learning_rate": 9.979097845289214e-06,
      "loss": 0.4841,
      "step": 388
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.359676480293274,
      "learning_rate": 9.978875047678407e-06,
      "loss": 0.4934,
      "step": 389
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9471995830535889,
      "learning_rate": 9.978651071459581e-06,
      "loss": 0.5367,
      "step": 390
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5461609363555908,
      "learning_rate": 9.978425916685758e-06,
      "loss": 0.4893,
      "step": 391
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7463409900665283,
      "learning_rate": 9.978199583410237e-06,
      "loss": 0.4563,
      "step": 392
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1404452323913574,
      "learning_rate": 9.977972071686596e-06,
      "loss": 0.4041,
      "step": 393
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.097142457962036,
      "learning_rate": 9.97774338156869e-06,
      "loss": 0.4887,
      "step": 394
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3016196489334106,
      "learning_rate": 9.977513513110656e-06,
      "loss": 0.5181,
      "step": 395
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1567115783691406,
      "learning_rate": 9.97728246636691e-06,
      "loss": 0.4792,
      "step": 396
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.399734377861023,
      "learning_rate": 9.977050241392145e-06,
      "loss": 0.4883,
      "step": 397
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6048067808151245,
      "learning_rate": 9.976816838241334e-06,
      "loss": 0.4582,
      "step": 398
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4002853631973267,
      "learning_rate": 9.976582256969726e-06,
      "loss": 0.4902,
      "step": 399
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3310648202896118,
      "learning_rate": 9.976346497632857e-06,
      "loss": 0.5237,
      "step": 400
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2730402946472168,
      "learning_rate": 9.976109560286532e-06,
      "loss": 0.5145,
      "step": 401
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5454782247543335,
      "learning_rate": 9.975871444986837e-06,
      "loss": 0.484,
      "step": 402
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4063576459884644,
      "learning_rate": 9.975632151790146e-06,
      "loss": 0.5409,
      "step": 403
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.302215337753296,
      "learning_rate": 9.9753916807531e-06,
      "loss": 0.5195,
      "step": 404
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3657333850860596,
      "learning_rate": 9.975150031932626e-06,
      "loss": 0.4518,
      "step": 405
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4494612216949463,
      "learning_rate": 9.974907205385924e-06,
      "loss": 0.4972,
      "step": 406
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3071740865707397,
      "learning_rate": 9.97466320117048e-06,
      "loss": 0.425,
      "step": 407
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7450742721557617,
      "learning_rate": 9.974418019344055e-06,
      "loss": 0.5168,
      "step": 408
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3033392429351807,
      "learning_rate": 9.974171659964688e-06,
      "loss": 0.3989,
      "step": 409
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4481068849563599,
      "learning_rate": 9.973924123090698e-06,
      "loss": 0.4594,
      "step": 410
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7059589624404907,
      "learning_rate": 9.97367540878068e-06,
      "loss": 0.5005,
      "step": 411
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3467029333114624,
      "learning_rate": 9.973425517093511e-06,
      "loss": 0.5407,
      "step": 412
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2696888446807861,
      "learning_rate": 9.973174448088347e-06,
      "loss": 0.5048,
      "step": 413
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6999872922897339,
      "learning_rate": 9.972922201824625e-06,
      "loss": 0.5673,
      "step": 414
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2097415924072266,
      "learning_rate": 9.97266877836205e-06,
      "loss": 0.5484,
      "step": 415
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6003798246383667,
      "learning_rate": 9.972414177760616e-06,
      "loss": 0.5058,
      "step": 416
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8849529027938843,
      "learning_rate": 9.972158400080594e-06,
      "loss": 0.4337,
      "step": 417
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0863606929779053,
      "learning_rate": 9.97190144538253e-06,
      "loss": 0.5057,
      "step": 418
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.4546401500701904,
      "learning_rate": 9.97164331372725e-06,
      "loss": 0.442,
      "step": 419
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.6911532878875732,
      "learning_rate": 9.971384005175863e-06,
      "loss": 0.4815,
      "step": 420
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5365067720413208,
      "learning_rate": 9.971123519789752e-06,
      "loss": 0.4461,
      "step": 421
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9719572067260742,
      "learning_rate": 9.970861857630578e-06,
      "loss": 0.4776,
      "step": 422
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2094950675964355,
      "learning_rate": 9.97059901876028e-06,
      "loss": 0.4662,
      "step": 423
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2703169584274292,
      "learning_rate": 9.970335003241084e-06,
      "loss": 0.5101,
      "step": 424
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.085042953491211,
      "learning_rate": 9.970069811135482e-06,
      "loss": 0.5642,
      "step": 425
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6713632345199585,
      "learning_rate": 9.969803442506253e-06,
      "loss": 0.49,
      "step": 426
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5551130771636963,
      "learning_rate": 9.969535897416457e-06,
      "loss": 0.5094,
      "step": 427
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3387330770492554,
      "learning_rate": 9.969267175929418e-06,
      "loss": 0.4981,
      "step": 428
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4794032573699951,
      "learning_rate": 9.968997278108756e-06,
      "loss": 0.4719,
      "step": 429
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.747295618057251,
      "learning_rate": 9.968726204018359e-06,
      "loss": 0.5497,
      "step": 430
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.237779140472412,
      "learning_rate": 9.968453953722394e-06,
      "loss": 0.7268,
      "step": 431
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.474793553352356,
      "learning_rate": 9.968180527285314e-06,
      "loss": 0.5277,
      "step": 432
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4511998891830444,
      "learning_rate": 9.96790592477184e-06,
      "loss": 0.4865,
      "step": 433
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.527146339416504,
      "learning_rate": 9.967630146246979e-06,
      "loss": 0.4535,
      "step": 434
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9981977343559265,
      "learning_rate": 9.967353191776013e-06,
      "loss": 0.7394,
      "step": 435
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.321199893951416,
      "learning_rate": 9.967075061424503e-06,
      "loss": 0.4447,
      "step": 436
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3631044626235962,
      "learning_rate": 9.96679575525829e-06,
      "loss": 0.4347,
      "step": 437
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8455913066864014,
      "learning_rate": 9.966515273343488e-06,
      "loss": 0.5395,
      "step": 438
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.885399341583252,
      "learning_rate": 9.966233615746496e-06,
      "loss": 0.5401,
      "step": 439
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3148796558380127,
      "learning_rate": 9.965950782533991e-06,
      "loss": 0.4434,
      "step": 440
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1720497608184814,
      "learning_rate": 9.965666773772919e-06,
      "loss": 0.5587,
      "step": 441
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4685440063476562,
      "learning_rate": 9.965381589530519e-06,
      "loss": 0.4898,
      "step": 442
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.177178144454956,
      "learning_rate": 9.965095229874293e-06,
      "loss": 0.7481,
      "step": 443
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4206444025039673,
      "learning_rate": 9.964807694872034e-06,
      "loss": 0.4966,
      "step": 444
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4452723264694214,
      "learning_rate": 9.964518984591804e-06,
      "loss": 0.5536,
      "step": 445
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9196152687072754,
      "learning_rate": 9.96422909910195e-06,
      "loss": 0.4718,
      "step": 446
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4781453609466553,
      "learning_rate": 9.963938038471091e-06,
      "loss": 0.4602,
      "step": 447
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2233330011367798,
      "learning_rate": 9.963645802768132e-06,
      "loss": 0.4751,
      "step": 448
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6934711933135986,
      "learning_rate": 9.963352392062246e-06,
      "loss": 0.5226,
      "step": 449
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.992238163948059,
      "learning_rate": 9.963057806422895e-06,
      "loss": 0.4631,
      "step": 450
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3681590557098389,
      "learning_rate": 9.96276204591981e-06,
      "loss": 0.7505,
      "step": 451
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7012890577316284,
      "learning_rate": 9.962465110623005e-06,
      "loss": 0.4867,
      "step": 452
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1665573120117188,
      "learning_rate": 9.962167000602771e-06,
      "loss": 0.5117,
      "step": 453
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0257365703582764,
      "learning_rate": 9.961867715929679e-06,
      "loss": 0.4813,
      "step": 454
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4214024543762207,
      "learning_rate": 9.961567256674574e-06,
      "loss": 0.5194,
      "step": 455
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4145443439483643,
      "learning_rate": 9.96126562290858e-06,
      "loss": 0.509,
      "step": 456
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.640019416809082,
      "learning_rate": 9.960962814703103e-06,
      "loss": 0.4765,
      "step": 457
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2582725286483765,
      "learning_rate": 9.960658832129826e-06,
      "loss": 0.4928,
      "step": 458
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3696701526641846,
      "learning_rate": 9.960353675260703e-06,
      "loss": 0.4761,
      "step": 459
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2243882417678833,
      "learning_rate": 9.960047344167974e-06,
      "loss": 0.407,
      "step": 460
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9993559122085571,
      "learning_rate": 9.959739838924156e-06,
      "loss": 0.4769,
      "step": 461
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3125697374343872,
      "learning_rate": 9.959431159602037e-06,
      "loss": 0.5171,
      "step": 462
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6583348512649536,
      "learning_rate": 9.959121306274695e-06,
      "loss": 0.5048,
      "step": 463
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5316861867904663,
      "learning_rate": 9.958810279015474e-06,
      "loss": 0.529,
      "step": 464
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7678556442260742,
      "learning_rate": 9.958498077898002e-06,
      "loss": 0.4931,
      "step": 465
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.253062129020691,
      "learning_rate": 9.958184702996186e-06,
      "loss": 0.4219,
      "step": 466
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4750593900680542,
      "learning_rate": 9.957870154384206e-06,
      "loss": 0.5051,
      "step": 467
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.313337802886963,
      "learning_rate": 9.957554432136523e-06,
      "loss": 0.5077,
      "step": 468
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5685169696807861,
      "learning_rate": 9.957237536327875e-06,
      "loss": 0.4738,
      "step": 469
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2730175256729126,
      "learning_rate": 9.95691946703328e-06,
      "loss": 0.4958,
      "step": 470
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1391661167144775,
      "learning_rate": 9.95660022432803e-06,
      "loss": 0.447,
      "step": 471
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0782504081726074,
      "learning_rate": 9.9562798082877e-06,
      "loss": 0.4889,
      "step": 472
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1859179735183716,
      "learning_rate": 9.955958218988133e-06,
      "loss": 0.4362,
      "step": 473
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.168246030807495,
      "learning_rate": 9.955635456505462e-06,
      "loss": 0.5354,
      "step": 474
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1046154499053955,
      "learning_rate": 9.95531152091609e-06,
      "loss": 0.5534,
      "step": 475
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5351957082748413,
      "learning_rate": 9.9549864122967e-06,
      "loss": 0.4745,
      "step": 476
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1956177949905396,
      "learning_rate": 9.95466013072425e-06,
      "loss": 0.7468,
      "step": 477
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.288083553314209,
      "learning_rate": 9.954332676275982e-06,
      "loss": 0.5199,
      "step": 478
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3561666011810303,
      "learning_rate": 9.954004049029409e-06,
      "loss": 0.7686,
      "step": 479
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1468263864517212,
      "learning_rate": 9.953674249062323e-06,
      "loss": 0.4185,
      "step": 480
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.586303472518921,
      "learning_rate": 9.953343276452798e-06,
      "loss": 0.4664,
      "step": 481
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1910285949707031,
      "learning_rate": 9.95301113127918e-06,
      "loss": 0.472,
      "step": 482
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3395411968231201,
      "learning_rate": 9.952677813620097e-06,
      "loss": 0.5061,
      "step": 483
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3746095895767212,
      "learning_rate": 9.95234332355445e-06,
      "loss": 0.5245,
      "step": 484
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2486451864242554,
      "learning_rate": 9.952007661161423e-06,
      "loss": 0.5227,
      "step": 485
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.298546314239502,
      "learning_rate": 9.95167082652047e-06,
      "loss": 0.7406,
      "step": 486
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2662547826766968,
      "learning_rate": 9.951332819711332e-06,
      "loss": 0.4954,
      "step": 487
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9688695669174194,
      "learning_rate": 9.950993640814022e-06,
      "loss": 0.5298,
      "step": 488
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4438159465789795,
      "learning_rate": 9.950653289908828e-06,
      "loss": 0.5758,
      "step": 489
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2391834259033203,
      "learning_rate": 9.950311767076322e-06,
      "loss": 0.4374,
      "step": 490
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5483145713806152,
      "learning_rate": 9.949969072397346e-06,
      "loss": 0.5003,
      "step": 491
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5766024589538574,
      "learning_rate": 9.949625205953026e-06,
      "loss": 0.5763,
      "step": 492
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6223018169403076,
      "learning_rate": 9.949280167824762e-06,
      "loss": 0.4954,
      "step": 493
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3649510145187378,
      "learning_rate": 9.948933958094231e-06,
      "loss": 0.5662,
      "step": 494
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.449120283126831,
      "learning_rate": 9.948586576843391e-06,
      "loss": 0.5055,
      "step": 495
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.46847665309906,
      "learning_rate": 9.948238024154472e-06,
      "loss": 0.4727,
      "step": 496
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.689347505569458,
      "learning_rate": 9.947888300109988e-06,
      "loss": 0.4333,
      "step": 497
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2207201719284058,
      "learning_rate": 9.94753740479272e-06,
      "loss": 0.4448,
      "step": 498
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3758023977279663,
      "learning_rate": 9.947185338285736e-06,
      "loss": 0.4356,
      "step": 499
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1571437120437622,
      "learning_rate": 9.94683210067238e-06,
      "loss": 0.4446,
      "step": 500
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4641218185424805,
      "learning_rate": 9.946477692036268e-06,
      "loss": 0.5206,
      "step": 501
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1962779760360718,
      "learning_rate": 9.946122112461296e-06,
      "loss": 0.5936,
      "step": 502
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9331399202346802,
      "learning_rate": 9.94576536203164e-06,
      "loss": 0.5214,
      "step": 503
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2111033201217651,
      "learning_rate": 9.945407440831747e-06,
      "loss": 0.5289,
      "step": 504
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4663385152816772,
      "learning_rate": 9.945048348946348e-06,
      "loss": 0.5059,
      "step": 505
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4874545335769653,
      "learning_rate": 9.944688086460446e-06,
      "loss": 0.4695,
      "step": 506
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.991346001625061,
      "learning_rate": 9.944326653459324e-06,
      "loss": 0.4994,
      "step": 507
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6647307872772217,
      "learning_rate": 9.94396405002854e-06,
      "loss": 0.4748,
      "step": 508
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2705403566360474,
      "learning_rate": 9.943600276253933e-06,
      "loss": 0.5481,
      "step": 509
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1293013095855713,
      "learning_rate": 9.943235332221613e-06,
      "loss": 0.4569,
      "step": 510
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3067750930786133,
      "learning_rate": 9.942869218017969e-06,
      "loss": 0.5195,
      "step": 511
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4827390909194946,
      "learning_rate": 9.94250193372967e-06,
      "loss": 0.575,
      "step": 512
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.958248496055603,
      "learning_rate": 9.942133479443665e-06,
      "loss": 0.527,
      "step": 513
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5360702276229858,
      "learning_rate": 9.941763855247167e-06,
      "loss": 0.5072,
      "step": 514
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.086057424545288,
      "learning_rate": 9.941393061227677e-06,
      "loss": 0.4152,
      "step": 515
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.138413667678833,
      "learning_rate": 9.941021097472973e-06,
      "loss": 0.4776,
      "step": 516
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2786697149276733,
      "learning_rate": 9.940647964071104e-06,
      "loss": 0.7657,
      "step": 517
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.187507390975952,
      "learning_rate": 9.9402736611104e-06,
      "loss": 0.4486,
      "step": 518
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6601505279541016,
      "learning_rate": 9.939898188679465e-06,
      "loss": 0.4761,
      "step": 519
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9516855478286743,
      "learning_rate": 9.939521546867182e-06,
      "loss": 0.4328,
      "step": 520
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8911758661270142,
      "learning_rate": 9.939143735762714e-06,
      "loss": 0.4832,
      "step": 521
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5340546369552612,
      "learning_rate": 9.93876475545549e-06,
      "loss": 0.4128,
      "step": 522
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.565895676612854,
      "learning_rate": 9.938384606035229e-06,
      "loss": 0.5128,
      "step": 523
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3527443408966064,
      "learning_rate": 9.938003287591918e-06,
      "loss": 0.5237,
      "step": 524
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4296656847000122,
      "learning_rate": 9.937620800215824e-06,
      "loss": 0.5061,
      "step": 525
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4053155183792114,
      "learning_rate": 9.93723714399749e-06,
      "loss": 0.4841,
      "step": 526
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.405720829963684,
      "learning_rate": 9.936852319027734e-06,
      "loss": 0.4905,
      "step": 527
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2800415754318237,
      "learning_rate": 9.936466325397657e-06,
      "loss": 0.5158,
      "step": 528
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.409533977508545,
      "learning_rate": 9.936079163198628e-06,
      "loss": 0.4328,
      "step": 529
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2527406215667725,
      "learning_rate": 9.935690832522299e-06,
      "loss": 0.4504,
      "step": 530
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1247234344482422,
      "learning_rate": 9.935301333460594e-06,
      "loss": 0.4364,
      "step": 531
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3514612913131714,
      "learning_rate": 9.934910666105716e-06,
      "loss": 0.4844,
      "step": 532
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2109562158584595,
      "learning_rate": 9.934518830550147e-06,
      "loss": 0.4624,
      "step": 533
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0861914157867432,
      "learning_rate": 9.934125826886642e-06,
      "loss": 0.537,
      "step": 534
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5671873092651367,
      "learning_rate": 9.933731655208232e-06,
      "loss": 0.4372,
      "step": 535
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.376004934310913,
      "learning_rate": 9.933336315608229e-06,
      "loss": 0.4642,
      "step": 536
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3953100442886353,
      "learning_rate": 9.932939808180215e-06,
      "loss": 0.4862,
      "step": 537
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4472007751464844,
      "learning_rate": 9.932542133018054e-06,
      "loss": 0.4668,
      "step": 538
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3800816535949707,
      "learning_rate": 9.932143290215885e-06,
      "loss": 0.502,
      "step": 539
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.684339165687561,
      "learning_rate": 9.93174327986812e-06,
      "loss": 0.4985,
      "step": 540
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1112492084503174,
      "learning_rate": 9.931342102069453e-06,
      "loss": 0.5418,
      "step": 541
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5445326566696167,
      "learning_rate": 9.93093975691485e-06,
      "loss": 0.4463,
      "step": 542
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.427912712097168,
      "learning_rate": 9.930536244499556e-06,
      "loss": 0.459,
      "step": 543
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2257673740386963,
      "learning_rate": 9.93013156491909e-06,
      "loss": 0.4697,
      "step": 544
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2631738185882568,
      "learning_rate": 9.929725718269248e-06,
      "loss": 0.5119,
      "step": 545
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8101071119308472,
      "learning_rate": 9.929318704646105e-06,
      "loss": 0.4798,
      "step": 546
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.594286561012268,
      "learning_rate": 9.928910524146007e-06,
      "loss": 0.5318,
      "step": 547
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0718897581100464,
      "learning_rate": 9.928501176865581e-06,
      "loss": 0.498,
      "step": 548
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2770321369171143,
      "learning_rate": 9.928090662901728e-06,
      "loss": 0.5387,
      "step": 549
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2339800596237183,
      "learning_rate": 9.927678982351627e-06,
      "loss": 0.4979,
      "step": 550
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6824342012405396,
      "learning_rate": 9.92726613531273e-06,
      "loss": 0.5311,
      "step": 551
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4239883422851562,
      "learning_rate": 9.926852121882766e-06,
      "loss": 0.5329,
      "step": 552
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3430242538452148,
      "learning_rate": 9.926436942159744e-06,
      "loss": 0.5111,
      "step": 553
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4747304916381836,
      "learning_rate": 9.926020596241942e-06,
      "loss": 0.4415,
      "step": 554
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2703022956848145,
      "learning_rate": 9.925603084227923e-06,
      "loss": 0.5324,
      "step": 555
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.313955545425415,
      "learning_rate": 9.92518440621652e-06,
      "loss": 0.5876,
      "step": 556
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0497421026229858,
      "learning_rate": 9.924764562306839e-06,
      "loss": 0.4874,
      "step": 557
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6241216659545898,
      "learning_rate": 9.92434355259827e-06,
      "loss": 0.516,
      "step": 558
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4948160648345947,
      "learning_rate": 9.923921377190476e-06,
      "loss": 0.4045,
      "step": 559
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1558606624603271,
      "learning_rate": 9.92349803618339e-06,
      "loss": 0.4228,
      "step": 560
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4027155637741089,
      "learning_rate": 9.923073529677232e-06,
      "loss": 0.5078,
      "step": 561
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2208970785140991,
      "learning_rate": 9.922647857772487e-06,
      "loss": 0.7308,
      "step": 562
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2489885091781616,
      "learning_rate": 9.922221020569924e-06,
      "loss": 0.4629,
      "step": 563
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.638601303100586,
      "learning_rate": 9.921793018170584e-06,
      "loss": 0.4335,
      "step": 564
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1841243505477905,
      "learning_rate": 9.921363850675785e-06,
      "loss": 0.5317,
      "step": 565
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2786054611206055,
      "learning_rate": 9.920933518187119e-06,
      "loss": 0.5064,
      "step": 566
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7184761762619019,
      "learning_rate": 9.920502020806456e-06,
      "loss": 0.4169,
      "step": 567
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9960412383079529,
      "learning_rate": 9.92006935863594e-06,
      "loss": 0.5114,
      "step": 568
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2250783443450928,
      "learning_rate": 9.91963553177799e-06,
      "loss": 0.4148,
      "step": 569
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2892866134643555,
      "learning_rate": 9.919200540335307e-06,
      "loss": 0.4796,
      "step": 570
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1436560153961182,
      "learning_rate": 9.918764384410859e-06,
      "loss": 0.5247,
      "step": 571
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7284610271453857,
      "learning_rate": 9.918327064107895e-06,
      "loss": 0.4526,
      "step": 572
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2348374128341675,
      "learning_rate": 9.917888579529937e-06,
      "loss": 0.4386,
      "step": 573
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1371278762817383,
      "learning_rate": 9.917448930780786e-06,
      "loss": 0.4813,
      "step": 574
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3099898099899292,
      "learning_rate": 9.917008117964515e-06,
      "loss": 0.7642,
      "step": 575
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.225331425666809,
      "learning_rate": 9.916566141185472e-06,
      "loss": 0.4732,
      "step": 576
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1973623037338257,
      "learning_rate": 9.916123000548287e-06,
      "loss": 0.4117,
      "step": 577
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6543819904327393,
      "learning_rate": 9.91567869615786e-06,
      "loss": 0.5377,
      "step": 578
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.37960946559906,
      "learning_rate": 9.915233228119364e-06,
      "loss": 0.456,
      "step": 579
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3884525299072266,
      "learning_rate": 9.914786596538254e-06,
      "loss": 0.4894,
      "step": 580
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9041696786880493,
      "learning_rate": 9.914338801520256e-06,
      "loss": 0.5,
      "step": 581
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4105557203292847,
      "learning_rate": 9.913889843171373e-06,
      "loss": 0.5634,
      "step": 582
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4263547658920288,
      "learning_rate": 9.913439721597885e-06,
      "loss": 0.5072,
      "step": 583
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0769346952438354,
      "learning_rate": 9.912988436906344e-06,
      "loss": 0.4572,
      "step": 584
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7723461389541626,
      "learning_rate": 9.912535989203578e-06,
      "loss": 0.4605,
      "step": 585
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.854954242706299,
      "learning_rate": 9.912082378596694e-06,
      "loss": 0.5417,
      "step": 586
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.137885332107544,
      "learning_rate": 9.911627605193068e-06,
      "loss": 0.4964,
      "step": 587
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2571381330490112,
      "learning_rate": 9.911171669100356e-06,
      "loss": 0.7681,
      "step": 588
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2001739740371704,
      "learning_rate": 9.910714570426491e-06,
      "loss": 0.5063,
      "step": 589
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3071500062942505,
      "learning_rate": 9.910256309279676e-06,
      "loss": 0.7429,
      "step": 590
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.49067223072052,
      "learning_rate": 9.90979688576839e-06,
      "loss": 0.5312,
      "step": 591
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.297112226486206,
      "learning_rate": 9.90933630000139e-06,
      "loss": 0.4694,
      "step": 592
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.527873039245605,
      "learning_rate": 9.908874552087704e-06,
      "loss": 0.467,
      "step": 593
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.577333927154541,
      "learning_rate": 9.908411642136643e-06,
      "loss": 0.5048,
      "step": 594
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2879811525344849,
      "learning_rate": 9.907947570257783e-06,
      "loss": 0.5941,
      "step": 595
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4200749397277832,
      "learning_rate": 9.907482336560985e-06,
      "loss": 0.4583,
      "step": 596
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1413110494613647,
      "learning_rate": 9.907015941156374e-06,
      "loss": 0.5298,
      "step": 597
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.663284420967102,
      "learning_rate": 9.90654838415436e-06,
      "loss": 0.5298,
      "step": 598
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5047131776809692,
      "learning_rate": 9.906079665665622e-06,
      "loss": 0.4325,
      "step": 599
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9759047627449036,
      "learning_rate": 9.905609785801115e-06,
      "loss": 0.4456,
      "step": 600
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3253341913223267,
      "learning_rate": 9.905138744672073e-06,
      "loss": 0.5287,
      "step": 601
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1192137002944946,
      "learning_rate": 9.904666542390001e-06,
      "loss": 0.4187,
      "step": 602
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.32114577293396,
      "learning_rate": 9.904193179066678e-06,
      "loss": 0.4137,
      "step": 603
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2756911516189575,
      "learning_rate": 9.90371865481416e-06,
      "loss": 0.5386,
      "step": 604
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5760791301727295,
      "learning_rate": 9.903242969744776e-06,
      "loss": 0.4841,
      "step": 605
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3466516733169556,
      "learning_rate": 9.902766123971132e-06,
      "loss": 0.421,
      "step": 606
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6907132863998413,
      "learning_rate": 9.902288117606109e-06,
      "loss": 0.4679,
      "step": 607
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5313516855239868,
      "learning_rate": 9.901808950762861e-06,
      "loss": 0.5182,
      "step": 608
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5687532424926758,
      "learning_rate": 9.901328623554815e-06,
      "loss": 0.4744,
      "step": 609
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7481404542922974,
      "learning_rate": 9.900847136095677e-06,
      "loss": 0.4748,
      "step": 610
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.4184763431549072,
      "learning_rate": 9.900364488499426e-06,
      "loss": 0.4189,
      "step": 611
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2914190292358398,
      "learning_rate": 9.899880680880315e-06,
      "loss": 0.5562,
      "step": 612
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.870982050895691,
      "learning_rate": 9.89939571335287e-06,
      "loss": 0.4911,
      "step": 613
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5423040390014648,
      "learning_rate": 9.898909586031893e-06,
      "loss": 0.541,
      "step": 614
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6213878393173218,
      "learning_rate": 9.898422299032465e-06,
      "loss": 0.4637,
      "step": 615
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1832079887390137,
      "learning_rate": 9.897933852469932e-06,
      "loss": 0.4291,
      "step": 616
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2968677282333374,
      "learning_rate": 9.897444246459925e-06,
      "loss": 0.5457,
      "step": 617
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3650062084197998,
      "learning_rate": 9.896953481118341e-06,
      "loss": 0.4644,
      "step": 618
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.369401454925537,
      "learning_rate": 9.896461556561356e-06,
      "loss": 0.5144,
      "step": 619
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2321596145629883,
      "learning_rate": 9.895968472905419e-06,
      "loss": 0.7423,
      "step": 620
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.084082007408142,
      "learning_rate": 9.895474230267254e-06,
      "loss": 0.4166,
      "step": 621
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4489797353744507,
      "learning_rate": 9.894978828763859e-06,
      "loss": 0.4885,
      "step": 622
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2688915729522705,
      "learning_rate": 9.894482268512506e-06,
      "loss": 0.4527,
      "step": 623
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1376956701278687,
      "learning_rate": 9.893984549630742e-06,
      "loss": 0.5501,
      "step": 624
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1803518533706665,
      "learning_rate": 9.893485672236387e-06,
      "loss": 0.4204,
      "step": 625
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3184322118759155,
      "learning_rate": 9.892985636447537e-06,
      "loss": 0.4761,
      "step": 626
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5341038703918457,
      "learning_rate": 9.89248444238256e-06,
      "loss": 0.4643,
      "step": 627
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1612651348114014,
      "learning_rate": 9.891982090160101e-06,
      "loss": 0.4846,
      "step": 628
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2269127368927002,
      "learning_rate": 9.89147857989908e-06,
      "loss": 0.4711,
      "step": 629
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.091698169708252,
      "learning_rate": 9.890973911718682e-06,
      "loss": 0.503,
      "step": 630
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3240936994552612,
      "learning_rate": 9.89046808573838e-06,
      "loss": 0.5525,
      "step": 631
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3879518508911133,
      "learning_rate": 9.889961102077909e-06,
      "loss": 0.563,
      "step": 632
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3189399242401123,
      "learning_rate": 9.889452960857287e-06,
      "loss": 0.5419,
      "step": 633
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.085101842880249,
      "learning_rate": 9.8889436621968e-06,
      "loss": 0.4816,
      "step": 634
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2582340240478516,
      "learning_rate": 9.88843320621701e-06,
      "loss": 0.5196,
      "step": 635
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.409833550453186,
      "learning_rate": 9.887921593038752e-06,
      "loss": 0.5051,
      "step": 636
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.387506365776062,
      "learning_rate": 9.887408822783137e-06,
      "loss": 0.4156,
      "step": 637
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4254401922225952,
      "learning_rate": 9.886894895571552e-06,
      "loss": 0.5581,
      "step": 638
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6934850215911865,
      "learning_rate": 9.88637981152565e-06,
      "loss": 0.466,
      "step": 639
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.243393898010254,
      "learning_rate": 9.885863570767365e-06,
      "loss": 0.5011,
      "step": 640
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.288864254951477,
      "learning_rate": 9.885346173418902e-06,
      "loss": 0.4051,
      "step": 641
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2027921676635742,
      "learning_rate": 9.884827619602739e-06,
      "loss": 0.5392,
      "step": 642
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4817938804626465,
      "learning_rate": 9.88430790944163e-06,
      "loss": 0.4241,
      "step": 643
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3582684993743896,
      "learning_rate": 9.883787043058604e-06,
      "loss": 0.4334,
      "step": 644
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2040777206420898,
      "learning_rate": 9.883265020576958e-06,
      "loss": 0.4579,
      "step": 645
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5883030891418457,
      "learning_rate": 9.882741842120266e-06,
      "loss": 0.4881,
      "step": 646
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1655983924865723,
      "learning_rate": 9.88221750781238e-06,
      "loss": 0.4756,
      "step": 647
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7614014148712158,
      "learning_rate": 9.881692017777416e-06,
      "loss": 0.5511,
      "step": 648
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3858851194381714,
      "learning_rate": 9.881165372139772e-06,
      "loss": 0.5441,
      "step": 649
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4027568101882935,
      "learning_rate": 9.880637571024115e-06,
      "loss": 0.4623,
      "step": 650
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.397531509399414,
      "learning_rate": 9.880108614555387e-06,
      "loss": 0.7829,
      "step": 651
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4233757257461548,
      "learning_rate": 9.879578502858806e-06,
      "loss": 0.5536,
      "step": 652
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.921164870262146,
      "learning_rate": 9.879047236059858e-06,
      "loss": 0.4882,
      "step": 653
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1227595806121826,
      "learning_rate": 9.878514814284308e-06,
      "loss": 0.4882,
      "step": 654
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2329504489898682,
      "learning_rate": 9.87798123765819e-06,
      "loss": 0.3737,
      "step": 655
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1991897821426392,
      "learning_rate": 9.877446506307812e-06,
      "loss": 0.5083,
      "step": 656
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2141222953796387,
      "learning_rate": 9.87691062035976e-06,
      "loss": 0.5175,
      "step": 657
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.367884874343872,
      "learning_rate": 9.876373579940886e-06,
      "loss": 0.5843,
      "step": 658
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1761689186096191,
      "learning_rate": 9.875835385178323e-06,
      "loss": 0.4899,
      "step": 659
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7011725902557373,
      "learning_rate": 9.875296036199472e-06,
      "loss": 0.4185,
      "step": 660
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4056432247161865,
      "learning_rate": 9.874755533132006e-06,
      "loss": 0.4878,
      "step": 661
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2498486042022705,
      "learning_rate": 9.874213876103877e-06,
      "loss": 0.4754,
      "step": 662
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3387404680252075,
      "learning_rate": 9.873671065243306e-06,
      "loss": 0.468,
      "step": 663
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.548264980316162,
      "learning_rate": 9.87312710067879e-06,
      "loss": 0.524,
      "step": 664
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2542803287506104,
      "learning_rate": 9.872581982539093e-06,
      "loss": 0.46,
      "step": 665
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4770750999450684,
      "learning_rate": 9.872035710953261e-06,
      "loss": 0.4399,
      "step": 666
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3275631666183472,
      "learning_rate": 9.871488286050603e-06,
      "loss": 0.5455,
      "step": 667
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2804017066955566,
      "learning_rate": 9.870939707960712e-06,
      "loss": 0.4908,
      "step": 668
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2510147094726562,
      "learning_rate": 9.870389976813444e-06,
      "loss": 0.5263,
      "step": 669
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.127665638923645,
      "learning_rate": 9.869839092738935e-06,
      "loss": 0.5206,
      "step": 670
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.289987325668335,
      "learning_rate": 9.869287055867591e-06,
      "loss": 0.4315,
      "step": 671
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5064423084259033,
      "learning_rate": 9.86873386633009e-06,
      "loss": 0.4945,
      "step": 672
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8994383811950684,
      "learning_rate": 9.868179524257384e-06,
      "loss": 0.5393,
      "step": 673
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2275041341781616,
      "learning_rate": 9.8676240297807e-06,
      "loss": 0.4873,
      "step": 674
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3892773389816284,
      "learning_rate": 9.867067383031532e-06,
      "loss": 0.475,
      "step": 675
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5412677526474,
      "learning_rate": 9.866509584141655e-06,
      "loss": 0.4825,
      "step": 676
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4178550243377686,
      "learning_rate": 9.865950633243107e-06,
      "loss": 0.51,
      "step": 677
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.472943663597107,
      "learning_rate": 9.865390530468207e-06,
      "loss": 0.5376,
      "step": 678
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6006156206130981,
      "learning_rate": 9.864829275949543e-06,
      "loss": 0.5168,
      "step": 679
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.269654393196106,
      "learning_rate": 9.864266869819977e-06,
      "loss": 0.7336,
      "step": 680
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1786381006240845,
      "learning_rate": 9.86370331221264e-06,
      "loss": 0.5197,
      "step": 681
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2995963096618652,
      "learning_rate": 9.86313860326094e-06,
      "loss": 0.4414,
      "step": 682
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2939372062683105,
      "learning_rate": 9.862572743098557e-06,
      "loss": 0.4557,
      "step": 683
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4177366495132446,
      "learning_rate": 9.862005731859443e-06,
      "loss": 0.4384,
      "step": 684
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2183218002319336,
      "learning_rate": 9.861437569677819e-06,
      "loss": 0.5075,
      "step": 685
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6949011087417603,
      "learning_rate": 9.860868256688181e-06,
      "loss": 0.5049,
      "step": 686
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3184292316436768,
      "learning_rate": 9.860297793025301e-06,
      "loss": 0.5689,
      "step": 687
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.752382516860962,
      "learning_rate": 9.859726178824218e-06,
      "loss": 0.5135,
      "step": 688
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1489673852920532,
      "learning_rate": 9.859153414220247e-06,
      "loss": 0.5156,
      "step": 689
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6907415390014648,
      "learning_rate": 9.858579499348972e-06,
      "loss": 0.5069,
      "step": 690
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4562503099441528,
      "learning_rate": 9.858004434346253e-06,
      "loss": 0.5153,
      "step": 691
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.8911352157592773,
      "learning_rate": 9.85742821934822e-06,
      "loss": 0.5366,
      "step": 692
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4188759326934814,
      "learning_rate": 9.856850854491273e-06,
      "loss": 0.5474,
      "step": 693
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5671916007995605,
      "learning_rate": 9.85627233991209e-06,
      "loss": 0.5106,
      "step": 694
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2536547183990479,
      "learning_rate": 9.855692675747616e-06,
      "loss": 0.4424,
      "step": 695
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3865230083465576,
      "learning_rate": 9.855111862135072e-06,
      "loss": 0.4105,
      "step": 696
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7633193731307983,
      "learning_rate": 9.854529899211948e-06,
      "loss": 0.483,
      "step": 697
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8375548124313354,
      "learning_rate": 9.853946787116006e-06,
      "loss": 0.4921,
      "step": 698
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2581480741500854,
      "learning_rate": 9.853362525985286e-06,
      "loss": 0.4928,
      "step": 699
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0854853391647339,
      "learning_rate": 9.85277711595809e-06,
      "loss": 0.5241,
      "step": 700
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5144070386886597,
      "learning_rate": 9.852190557172999e-06,
      "loss": 0.5719,
      "step": 701
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3040661811828613,
      "learning_rate": 9.851602849768867e-06,
      "loss": 0.5003,
      "step": 702
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2903071641921997,
      "learning_rate": 9.851013993884813e-06,
      "loss": 0.4974,
      "step": 703
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.163820743560791,
      "learning_rate": 9.850423989660234e-06,
      "loss": 0.7307,
      "step": 704
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.446948766708374,
      "learning_rate": 9.849832837234796e-06,
      "loss": 0.5086,
      "step": 705
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1843669414520264,
      "learning_rate": 9.84924053674844e-06,
      "loss": 0.7506,
      "step": 706
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.329107403755188,
      "learning_rate": 9.848647088341372e-06,
      "loss": 0.5136,
      "step": 707
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.336033582687378,
      "learning_rate": 9.84805249215408e-06,
      "loss": 0.5523,
      "step": 708
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4298712015151978,
      "learning_rate": 9.847456748327312e-06,
      "loss": 0.5299,
      "step": 709
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3952280282974243,
      "learning_rate": 9.846859857002096e-06,
      "loss": 0.4816,
      "step": 710
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6183602809906006,
      "learning_rate": 9.84626181831973e-06,
      "loss": 0.4377,
      "step": 711
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.351163625717163,
      "learning_rate": 9.845662632421782e-06,
      "loss": 0.4724,
      "step": 712
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2861523628234863,
      "learning_rate": 9.845062299450094e-06,
      "loss": 0.5007,
      "step": 713
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6637217998504639,
      "learning_rate": 9.844460819546775e-06,
      "loss": 0.5374,
      "step": 714
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1165627241134644,
      "learning_rate": 9.84385819285421e-06,
      "loss": 0.459,
      "step": 715
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8170536756515503,
      "learning_rate": 9.843254419515053e-06,
      "loss": 0.5415,
      "step": 716
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3664236068725586,
      "learning_rate": 9.84264949967223e-06,
      "loss": 0.5532,
      "step": 717
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7114217281341553,
      "learning_rate": 9.842043433468942e-06,
      "loss": 0.5213,
      "step": 718
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2226399183273315,
      "learning_rate": 9.841436221048654e-06,
      "loss": 0.4978,
      "step": 719
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3316445350646973,
      "learning_rate": 9.840827862555107e-06,
      "loss": 0.4775,
      "step": 720
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3116806745529175,
      "learning_rate": 9.840218358132315e-06,
      "loss": 0.4996,
      "step": 721
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5289137363433838,
      "learning_rate": 9.83960770792456e-06,
      "loss": 0.5638,
      "step": 722
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.358001947402954,
      "learning_rate": 9.838995912076395e-06,
      "loss": 0.5223,
      "step": 723
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4830869436264038,
      "learning_rate": 9.838382970732646e-06,
      "loss": 0.5245,
      "step": 724
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6400556564331055,
      "learning_rate": 9.837768884038409e-06,
      "loss": 0.5422,
      "step": 725
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7164692878723145,
      "learning_rate": 9.837153652139053e-06,
      "loss": 0.5076,
      "step": 726
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4941353797912598,
      "learning_rate": 9.836537275180216e-06,
      "loss": 0.5705,
      "step": 727
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1875156164169312,
      "learning_rate": 9.835919753307808e-06,
      "loss": 0.7322,
      "step": 728
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3390061855316162,
      "learning_rate": 9.83530108666801e-06,
      "loss": 0.4778,
      "step": 729
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0525400638580322,
      "learning_rate": 9.834681275407275e-06,
      "loss": 0.4766,
      "step": 730
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2585258483886719,
      "learning_rate": 9.834060319672323e-06,
      "loss": 0.4484,
      "step": 731
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1731237173080444,
      "learning_rate": 9.833438219610148e-06,
      "loss": 0.716,
      "step": 732
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0599833726882935,
      "learning_rate": 9.832814975368017e-06,
      "loss": 0.4258,
      "step": 733
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3309118747711182,
      "learning_rate": 9.832190587093464e-06,
      "loss": 0.4034,
      "step": 734
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1239783763885498,
      "learning_rate": 9.831565054934297e-06,
      "loss": 0.4771,
      "step": 735
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7091649770736694,
      "learning_rate": 9.83093837903859e-06,
      "loss": 0.4902,
      "step": 736
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3880975246429443,
      "learning_rate": 9.830310559554694e-06,
      "loss": 0.4719,
      "step": 737
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7311679124832153,
      "learning_rate": 9.829681596631224e-06,
      "loss": 0.5411,
      "step": 738
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5941553115844727,
      "learning_rate": 9.829051490417074e-06,
      "loss": 0.4966,
      "step": 739
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0587382316589355,
      "learning_rate": 9.828420241061398e-06,
      "loss": 0.4996,
      "step": 740
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.766564130783081,
      "learning_rate": 9.82778784871363e-06,
      "loss": 0.478,
      "step": 741
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6132386922836304,
      "learning_rate": 9.827154313523473e-06,
      "loss": 0.4993,
      "step": 742
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3047584295272827,
      "learning_rate": 9.826519635640895e-06,
      "loss": 0.5544,
      "step": 743
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4630125761032104,
      "learning_rate": 9.825883815216137e-06,
      "loss": 0.4982,
      "step": 744
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2265208959579468,
      "learning_rate": 9.825246852399715e-06,
      "loss": 0.3906,
      "step": 745
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6273784637451172,
      "learning_rate": 9.82460874734241e-06,
      "loss": 0.4618,
      "step": 746
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.991787314414978,
      "learning_rate": 9.823969500195278e-06,
      "loss": 0.4602,
      "step": 747
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4785202741622925,
      "learning_rate": 9.823329111109639e-06,
      "loss": 0.4992,
      "step": 748
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.263999342918396,
      "learning_rate": 9.822687580237089e-06,
      "loss": 0.4934,
      "step": 749
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5625935792922974,
      "learning_rate": 9.82204490772949e-06,
      "loss": 0.5081,
      "step": 750
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2499037981033325,
      "learning_rate": 9.821401093738983e-06,
      "loss": 0.4938,
      "step": 751
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3405081033706665,
      "learning_rate": 9.820756138417966e-06,
      "loss": 0.461,
      "step": 752
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2549688816070557,
      "learning_rate": 9.820110041919115e-06,
      "loss": 0.4544,
      "step": 753
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1207332611083984,
      "learning_rate": 9.819462804395378e-06,
      "loss": 0.4282,
      "step": 754
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.475717544555664,
      "learning_rate": 9.818814425999967e-06,
      "loss": 0.4573,
      "step": 755
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1921522617340088,
      "learning_rate": 9.818164906886372e-06,
      "loss": 0.5104,
      "step": 756
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1879914999008179,
      "learning_rate": 9.817514247208341e-06,
      "loss": 0.4377,
      "step": 757
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5568978786468506,
      "learning_rate": 9.816862447119907e-06,
      "loss": 0.4313,
      "step": 758
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2427313327789307,
      "learning_rate": 9.816209506775361e-06,
      "loss": 0.5064,
      "step": 759
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1332916021347046,
      "learning_rate": 9.815555426329267e-06,
      "loss": 0.5585,
      "step": 760
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.487687110900879,
      "learning_rate": 9.814900205936464e-06,
      "loss": 0.4946,
      "step": 761
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1760735511779785,
      "learning_rate": 9.814243845752053e-06,
      "loss": 0.4048,
      "step": 762
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.226630687713623,
      "learning_rate": 9.813586345931412e-06,
      "loss": 0.5725,
      "step": 763
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3916456699371338,
      "learning_rate": 9.812927706630183e-06,
      "loss": 0.5328,
      "step": 764
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5703132152557373,
      "learning_rate": 9.812267928004282e-06,
      "loss": 0.44,
      "step": 765
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2705549001693726,
      "learning_rate": 9.811607010209893e-06,
      "loss": 0.4179,
      "step": 766
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0549081563949585,
      "learning_rate": 9.810944953403467e-06,
      "loss": 0.448,
      "step": 767
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.632178783416748,
      "learning_rate": 9.81028175774173e-06,
      "loss": 0.4686,
      "step": 768
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4825912714004517,
      "learning_rate": 9.809617423381677e-06,
      "loss": 0.4862,
      "step": 769
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8439841270446777,
      "learning_rate": 9.808951950480562e-06,
      "loss": 0.5202,
      "step": 770
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6393166780471802,
      "learning_rate": 9.808285339195927e-06,
      "loss": 0.4884,
      "step": 771
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4695415496826172,
      "learning_rate": 9.807617589685568e-06,
      "loss": 0.5013,
      "step": 772
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2011692523956299,
      "learning_rate": 9.806948702107555e-06,
      "loss": 0.722,
      "step": 773
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3457993268966675,
      "learning_rate": 9.80627867662023e-06,
      "loss": 0.4848,
      "step": 774
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5187811851501465,
      "learning_rate": 9.805607513382203e-06,
      "loss": 0.5009,
      "step": 775
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1142405271530151,
      "learning_rate": 9.804935212552351e-06,
      "loss": 0.5392,
      "step": 776
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.293678641319275,
      "learning_rate": 9.804261774289824e-06,
      "loss": 0.7564,
      "step": 777
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.34571373462677,
      "learning_rate": 9.80358719875404e-06,
      "loss": 0.478,
      "step": 778
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4421226978302002,
      "learning_rate": 9.802911486104684e-06,
      "loss": 0.5025,
      "step": 779
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2048964500427246,
      "learning_rate": 9.802234636501713e-06,
      "loss": 0.5374,
      "step": 780
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9232668280601501,
      "learning_rate": 9.80155665010535e-06,
      "loss": 0.5207,
      "step": 781
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.182893991470337,
      "learning_rate": 9.800877527076094e-06,
      "loss": 0.5572,
      "step": 782
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2248495817184448,
      "learning_rate": 9.800197267574703e-06,
      "loss": 0.4484,
      "step": 783
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.000931262969971,
      "learning_rate": 9.79951587176221e-06,
      "loss": 0.445,
      "step": 784
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3637051582336426,
      "learning_rate": 9.79883333979992e-06,
      "loss": 0.5082,
      "step": 785
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8909103870391846,
      "learning_rate": 9.7981496718494e-06,
      "loss": 0.5092,
      "step": 786
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.259360671043396,
      "learning_rate": 9.797464868072489e-06,
      "loss": 0.5539,
      "step": 787
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6029287576675415,
      "learning_rate": 9.796778928631294e-06,
      "loss": 0.516,
      "step": 788
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.853785753250122,
      "learning_rate": 9.796091853688195e-06,
      "loss": 0.5097,
      "step": 789
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5087240934371948,
      "learning_rate": 9.795403643405836e-06,
      "loss": 0.5042,
      "step": 790
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3299251794815063,
      "learning_rate": 9.79471429794713e-06,
      "loss": 0.4319,
      "step": 791
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.21416175365448,
      "learning_rate": 9.794023817475263e-06,
      "loss": 0.742,
      "step": 792
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1959669589996338,
      "learning_rate": 9.793332202153685e-06,
      "loss": 0.4776,
      "step": 793
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8605127334594727,
      "learning_rate": 9.792639452146116e-06,
      "loss": 0.4871,
      "step": 794
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.265860915184021,
      "learning_rate": 9.791945567616546e-06,
      "loss": 0.4919,
      "step": 795
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1639360189437866,
      "learning_rate": 9.791250548729231e-06,
      "loss": 0.4835,
      "step": 796
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.987308382987976,
      "learning_rate": 9.7905543956487e-06,
      "loss": 0.4351,
      "step": 797
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.080297589302063,
      "learning_rate": 9.789857108539742e-06,
      "loss": 0.3725,
      "step": 798
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2910720109939575,
      "learning_rate": 9.789158687567427e-06,
      "loss": 0.5013,
      "step": 799
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7998039722442627,
      "learning_rate": 9.788459132897083e-06,
      "loss": 0.4868,
      "step": 800
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5467863082885742,
      "learning_rate": 9.787758444694308e-06,
      "loss": 0.4916,
      "step": 801
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.360449194908142,
      "learning_rate": 9.787056623124973e-06,
      "loss": 0.526,
      "step": 802
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6561731100082397,
      "learning_rate": 9.786353668355214e-06,
      "loss": 0.4394,
      "step": 803
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6365082263946533,
      "learning_rate": 9.785649580551434e-06,
      "loss": 0.5144,
      "step": 804
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2892553806304932,
      "learning_rate": 9.78494435988031e-06,
      "loss": 0.4584,
      "step": 805
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4063256978988647,
      "learning_rate": 9.784238006508778e-06,
      "loss": 0.5053,
      "step": 806
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8231298923492432,
      "learning_rate": 9.78353052060405e-06,
      "loss": 0.4686,
      "step": 807
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.296527624130249,
      "learning_rate": 9.782821902333602e-06,
      "loss": 0.4942,
      "step": 808
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3531179428100586,
      "learning_rate": 9.782112151865181e-06,
      "loss": 0.5765,
      "step": 809
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7503249645233154,
      "learning_rate": 9.7814012693668e-06,
      "loss": 0.5651,
      "step": 810
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2552039623260498,
      "learning_rate": 9.780689255006737e-06,
      "loss": 0.5136,
      "step": 811
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1080585718154907,
      "learning_rate": 9.779976108953545e-06,
      "loss": 0.4326,
      "step": 812
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1324912309646606,
      "learning_rate": 9.779261831376043e-06,
      "loss": 0.5195,
      "step": 813
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1437087059020996,
      "learning_rate": 9.77854642244331e-06,
      "loss": 0.5445,
      "step": 814
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2919329404830933,
      "learning_rate": 9.777829882324705e-06,
      "loss": 0.4783,
      "step": 815
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1788330078125,
      "learning_rate": 9.777112211189843e-06,
      "loss": 0.5136,
      "step": 816
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.038795232772827,
      "learning_rate": 9.776393409208618e-06,
      "loss": 0.4178,
      "step": 817
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2931028604507446,
      "learning_rate": 9.775673476551181e-06,
      "loss": 0.4292,
      "step": 818
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0066335201263428,
      "learning_rate": 9.774952413387958e-06,
      "loss": 0.3935,
      "step": 819
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.283607840538025,
      "learning_rate": 9.77423021988964e-06,
      "loss": 0.4647,
      "step": 820
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0771129131317139,
      "learning_rate": 9.773506896227188e-06,
      "loss": 0.4718,
      "step": 821
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0620924234390259,
      "learning_rate": 9.772782442571825e-06,
      "loss": 0.4953,
      "step": 822
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3685996532440186,
      "learning_rate": 9.772056859095048e-06,
      "loss": 0.7609,
      "step": 823
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1477230787277222,
      "learning_rate": 9.771330145968616e-06,
      "loss": 0.7538,
      "step": 824
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.132218837738037,
      "learning_rate": 9.770602303364558e-06,
      "loss": 0.4782,
      "step": 825
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4066842794418335,
      "learning_rate": 9.769873331455172e-06,
      "loss": 0.4438,
      "step": 826
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.302470088005066,
      "learning_rate": 9.769143230413021e-06,
      "loss": 0.5593,
      "step": 827
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.295041561126709,
      "learning_rate": 9.768412000410934e-06,
      "loss": 0.4627,
      "step": 828
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.190081000328064,
      "learning_rate": 9.76767964162201e-06,
      "loss": 0.485,
      "step": 829
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3169193267822266,
      "learning_rate": 9.766946154219616e-06,
      "loss": 0.5484,
      "step": 830
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2089747190475464,
      "learning_rate": 9.766211538377382e-06,
      "loss": 0.476,
      "step": 831
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1917176246643066,
      "learning_rate": 9.765475794269208e-06,
      "loss": 0.4622,
      "step": 832
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4266786575317383,
      "learning_rate": 9.76473892206926e-06,
      "loss": 0.4781,
      "step": 833
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1548399925231934,
      "learning_rate": 9.764000921951976e-06,
      "loss": 0.4797,
      "step": 834
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8542743921279907,
      "learning_rate": 9.763261794092052e-06,
      "loss": 0.5073,
      "step": 835
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2357486486434937,
      "learning_rate": 9.762521538664458e-06,
      "loss": 0.4679,
      "step": 836
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2029500007629395,
      "learning_rate": 9.761780155844425e-06,
      "loss": 0.4584,
      "step": 837
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.2508387565612793,
      "learning_rate": 9.76103764580746e-06,
      "loss": 0.5015,
      "step": 838
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.456960916519165,
      "learning_rate": 9.760294008729327e-06,
      "loss": 0.5688,
      "step": 839
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2917060852050781,
      "learning_rate": 9.759549244786063e-06,
      "loss": 0.422,
      "step": 840
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3924790620803833,
      "learning_rate": 9.758803354153969e-06,
      "loss": 0.4515,
      "step": 841
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4664536714553833,
      "learning_rate": 9.758056337009614e-06,
      "loss": 0.5196,
      "step": 842
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3669981956481934,
      "learning_rate": 9.757308193529833e-06,
      "loss": 0.7554,
      "step": 843
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.906420350074768,
      "learning_rate": 9.756558923891728e-06,
      "loss": 0.5047,
      "step": 844
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5661225318908691,
      "learning_rate": 9.755808528272668e-06,
      "loss": 0.5398,
      "step": 845
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5264109373092651,
      "learning_rate": 9.755057006850286e-06,
      "loss": 0.6087,
      "step": 846
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.000998020172119,
      "learning_rate": 9.754304359802484e-06,
      "loss": 0.4593,
      "step": 847
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.347241759300232,
      "learning_rate": 9.753550587307432e-06,
      "loss": 0.4807,
      "step": 848
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.089613199234009,
      "learning_rate": 9.752795689543563e-06,
      "loss": 0.4239,
      "step": 849
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.291779041290283,
      "learning_rate": 9.752039666689578e-06,
      "loss": 0.4921,
      "step": 850
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8679808378219604,
      "learning_rate": 9.751282518924444e-06,
      "loss": 0.4958,
      "step": 851
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2798763513565063,
      "learning_rate": 9.750524246427392e-06,
      "loss": 0.4861,
      "step": 852
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3336844444274902,
      "learning_rate": 9.749764849377925e-06,
      "loss": 0.5126,
      "step": 853
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2998038530349731,
      "learning_rate": 9.749004327955808e-06,
      "loss": 0.4995,
      "step": 854
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.389527678489685,
      "learning_rate": 9.748242682341071e-06,
      "loss": 0.4529,
      "step": 855
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.484485387802124,
      "learning_rate": 9.747479912714015e-06,
      "loss": 0.5307,
      "step": 856
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3383790254592896,
      "learning_rate": 9.746716019255202e-06,
      "loss": 0.5308,
      "step": 857
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8206124305725098,
      "learning_rate": 9.745951002145464e-06,
      "loss": 0.5151,
      "step": 858
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3717045783996582,
      "learning_rate": 9.745184861565895e-06,
      "loss": 0.4438,
      "step": 859
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3200217485427856,
      "learning_rate": 9.74441759769786e-06,
      "loss": 0.449,
      "step": 860
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7551567554473877,
      "learning_rate": 9.743649210722983e-06,
      "loss": 0.5519,
      "step": 861
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.354387879371643,
      "learning_rate": 9.74287970082316e-06,
      "loss": 0.765,
      "step": 862
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5474889278411865,
      "learning_rate": 9.742109068180552e-06,
      "loss": 0.5524,
      "step": 863
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2600213289260864,
      "learning_rate": 9.741337312977584e-06,
      "loss": 0.5456,
      "step": 864
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0704703330993652,
      "learning_rate": 9.740564435396947e-06,
      "loss": 0.5146,
      "step": 865
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7180339097976685,
      "learning_rate": 9.739790435621595e-06,
      "loss": 0.4809,
      "step": 866
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0978929996490479,
      "learning_rate": 9.739015313834755e-06,
      "loss": 0.7386,
      "step": 867
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.245771884918213,
      "learning_rate": 9.738239070219913e-06,
      "loss": 0.4534,
      "step": 868
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.743398904800415,
      "learning_rate": 9.737461704960824e-06,
      "loss": 0.5314,
      "step": 869
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2402794361114502,
      "learning_rate": 9.736683218241505e-06,
      "loss": 0.512,
      "step": 870
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.746849775314331,
      "learning_rate": 9.735903610246243e-06,
      "loss": 0.4727,
      "step": 871
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2162542343139648,
      "learning_rate": 9.735122881159587e-06,
      "loss": 0.5341,
      "step": 872
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3153784275054932,
      "learning_rate": 9.734341031166355e-06,
      "loss": 0.5428,
      "step": 873
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9523868560791016,
      "learning_rate": 9.733558060451623e-06,
      "loss": 0.5878,
      "step": 874
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7059558629989624,
      "learning_rate": 9.73277396920074e-06,
      "loss": 0.4916,
      "step": 875
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.1047327518463135,
      "learning_rate": 9.731988757599317e-06,
      "loss": 0.5005,
      "step": 876
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5254147052764893,
      "learning_rate": 9.731202425833234e-06,
      "loss": 0.4849,
      "step": 877
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3477200269699097,
      "learning_rate": 9.730414974088627e-06,
      "loss": 0.4504,
      "step": 878
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6488879919052124,
      "learning_rate": 9.729626402551909e-06,
      "loss": 0.539,
      "step": 879
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0081851482391357,
      "learning_rate": 9.728836711409746e-06,
      "loss": 0.5753,
      "step": 880
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6909116506576538,
      "learning_rate": 9.72804590084908e-06,
      "loss": 0.5028,
      "step": 881
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7305270433425903,
      "learning_rate": 9.72725397105711e-06,
      "loss": 0.5416,
      "step": 882
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3701854944229126,
      "learning_rate": 9.726460922221302e-06,
      "loss": 0.4691,
      "step": 883
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2592328786849976,
      "learning_rate": 9.725666754529393e-06,
      "loss": 0.5073,
      "step": 884
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6072139739990234,
      "learning_rate": 9.724871468169375e-06,
      "loss": 0.5032,
      "step": 885
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4037764072418213,
      "learning_rate": 9.724075063329512e-06,
      "loss": 0.4504,
      "step": 886
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5726760625839233,
      "learning_rate": 9.723277540198328e-06,
      "loss": 0.4586,
      "step": 887
    },
    {
      "epoch": 0.13,
      "grad_norm": 11.143845558166504,
      "learning_rate": 9.722478898964617e-06,
      "loss": 0.3734,
      "step": 888
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0728116035461426,
      "learning_rate": 9.721679139817434e-06,
      "loss": 0.4067,
      "step": 889
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6497911214828491,
      "learning_rate": 9.720878262946098e-06,
      "loss": 0.4292,
      "step": 890
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.1607472896575928,
      "learning_rate": 9.720076268540198e-06,
      "loss": 0.4769,
      "step": 891
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2998348474502563,
      "learning_rate": 9.719273156789576e-06,
      "loss": 0.5051,
      "step": 892
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3718366622924805,
      "learning_rate": 9.718468927884352e-06,
      "loss": 0.3931,
      "step": 893
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1141403913497925,
      "learning_rate": 9.717663582014902e-06,
      "loss": 0.4856,
      "step": 894
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.188766598701477,
      "learning_rate": 9.716857119371871e-06,
      "loss": 0.433,
      "step": 895
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3458236455917358,
      "learning_rate": 9.716049540146164e-06,
      "loss": 0.5082,
      "step": 896
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6286412477493286,
      "learning_rate": 9.715240844528954e-06,
      "loss": 0.4723,
      "step": 897
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2093448638916016,
      "learning_rate": 9.714431032711678e-06,
      "loss": 0.5248,
      "step": 898
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.535269021987915,
      "learning_rate": 9.713620104886033e-06,
      "loss": 0.4947,
      "step": 899
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1516129970550537,
      "learning_rate": 9.712808061243984e-06,
      "loss": 0.7129,
      "step": 900
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4093316793441772,
      "learning_rate": 9.71199490197776e-06,
      "loss": 0.5122,
      "step": 901
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6282235383987427,
      "learning_rate": 9.711180627279854e-06,
      "loss": 0.5116,
      "step": 902
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.385067105293274,
      "learning_rate": 9.710365237343023e-06,
      "loss": 0.4462,
      "step": 903
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4457851648330688,
      "learning_rate": 9.709548732360286e-06,
      "loss": 0.5139,
      "step": 904
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8186821937561035,
      "learning_rate": 9.708731112524928e-06,
      "loss": 0.5201,
      "step": 905
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4928468465805054,
      "learning_rate": 9.707912378030498e-06,
      "loss": 0.459,
      "step": 906
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6225285530090332,
      "learning_rate": 9.707092529070808e-06,
      "loss": 0.4655,
      "step": 907
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4006320238113403,
      "learning_rate": 9.706271565839931e-06,
      "loss": 0.477,
      "step": 908
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0784704685211182,
      "learning_rate": 9.705449488532213e-06,
      "loss": 0.5241,
      "step": 909
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.016447901725769,
      "learning_rate": 9.704626297342254e-06,
      "loss": 0.398,
      "step": 910
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4202220439910889,
      "learning_rate": 9.703801992464921e-06,
      "loss": 0.441,
      "step": 911
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4927606582641602,
      "learning_rate": 9.702976574095345e-06,
      "loss": 0.4436,
      "step": 912
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.881305694580078,
      "learning_rate": 9.702150042428922e-06,
      "loss": 0.5213,
      "step": 913
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1095389127731323,
      "learning_rate": 9.701322397661308e-06,
      "loss": 0.4495,
      "step": 914
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.469618558883667,
      "learning_rate": 9.700493639988424e-06,
      "loss": 0.4992,
      "step": 915
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3795336484909058,
      "learning_rate": 9.69966376960646e-06,
      "loss": 0.4306,
      "step": 916
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7452293634414673,
      "learning_rate": 9.698832786711859e-06,
      "loss": 0.464,
      "step": 917
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3491458892822266,
      "learning_rate": 9.698000691501333e-06,
      "loss": 0.4401,
      "step": 918
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.382650375366211,
      "learning_rate": 9.697167484171862e-06,
      "loss": 0.5618,
      "step": 919
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.603818655014038,
      "learning_rate": 9.696333164920677e-06,
      "loss": 0.4897,
      "step": 920
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4779835939407349,
      "learning_rate": 9.695497733945286e-06,
      "loss": 0.5204,
      "step": 921
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2971168756484985,
      "learning_rate": 9.694661191443451e-06,
      "loss": 0.4885,
      "step": 922
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.516114592552185,
      "learning_rate": 9.693823537613199e-06,
      "loss": 0.4422,
      "step": 923
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2644919157028198,
      "learning_rate": 9.692984772652822e-06,
      "loss": 0.7617,
      "step": 924
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.438543677330017,
      "learning_rate": 9.692144896760873e-06,
      "loss": 0.4085,
      "step": 925
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2797496318817139,
      "learning_rate": 9.691303910136171e-06,
      "loss": 0.4771,
      "step": 926
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4143962860107422,
      "learning_rate": 9.690461812977794e-06,
      "loss": 0.4482,
      "step": 927
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9862637519836426,
      "learning_rate": 9.689618605485086e-06,
      "loss": 0.5203,
      "step": 928
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.397538661956787,
      "learning_rate": 9.68877428785765e-06,
      "loss": 0.5309,
      "step": 929
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6767852306365967,
      "learning_rate": 9.687928860295356e-06,
      "loss": 0.4998,
      "step": 930
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6708176136016846,
      "learning_rate": 9.687082322998335e-06,
      "loss": 0.423,
      "step": 931
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6972119808197021,
      "learning_rate": 9.686234676166983e-06,
      "loss": 0.5145,
      "step": 932
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.703012466430664,
      "learning_rate": 9.685385920001952e-06,
      "loss": 0.4733,
      "step": 933
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.148052453994751,
      "learning_rate": 9.684536054704165e-06,
      "loss": 0.4766,
      "step": 934
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2682818174362183,
      "learning_rate": 9.6836850804748e-06,
      "loss": 0.4381,
      "step": 935
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.6156415939331055,
      "learning_rate": 9.682832997515305e-06,
      "loss": 0.4927,
      "step": 936
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2621688842773438,
      "learning_rate": 9.681979806027383e-06,
      "loss": 0.5281,
      "step": 937
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4332431554794312,
      "learning_rate": 9.681125506213006e-06,
      "loss": 0.4961,
      "step": 938
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2000317573547363,
      "learning_rate": 9.680270098274403e-06,
      "loss": 0.4973,
      "step": 939
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.117930293083191,
      "learning_rate": 9.67941358241407e-06,
      "loss": 0.4321,
      "step": 940
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2645041942596436,
      "learning_rate": 9.678555958834762e-06,
      "loss": 0.4464,
      "step": 941
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4568620920181274,
      "learning_rate": 9.677697227739494e-06,
      "loss": 0.5101,
      "step": 942
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2090160846710205,
      "learning_rate": 9.67683738933155e-06,
      "loss": 0.4382,
      "step": 943
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0427026748657227,
      "learning_rate": 9.675976443814473e-06,
      "loss": 0.5099,
      "step": 944
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2095211744308472,
      "learning_rate": 9.675114391392065e-06,
      "loss": 0.5581,
      "step": 945
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2777276039123535,
      "learning_rate": 9.674251232268393e-06,
      "loss": 0.5253,
      "step": 946
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1630840301513672,
      "learning_rate": 9.673386966647787e-06,
      "loss": 0.4591,
      "step": 947
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8396165370941162,
      "learning_rate": 9.672521594734838e-06,
      "loss": 0.4499,
      "step": 948
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3729894161224365,
      "learning_rate": 9.671655116734396e-06,
      "loss": 0.5239,
      "step": 949
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2890291213989258,
      "learning_rate": 9.670787532851575e-06,
      "loss": 0.4179,
      "step": 950
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4093915224075317,
      "learning_rate": 9.669918843291753e-06,
      "loss": 0.4751,
      "step": 951
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4017232656478882,
      "learning_rate": 9.669049048260565e-06,
      "loss": 0.5725,
      "step": 952
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5517641305923462,
      "learning_rate": 9.668178147963915e-06,
      "loss": 0.4356,
      "step": 953
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5122113227844238,
      "learning_rate": 9.667306142607959e-06,
      "loss": 0.4495,
      "step": 954
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2669342756271362,
      "learning_rate": 9.666433032399123e-06,
      "loss": 0.4987,
      "step": 955
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0349496603012085,
      "learning_rate": 9.665558817544091e-06,
      "loss": 0.5306,
      "step": 956
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.475546956062317,
      "learning_rate": 9.664683498249807e-06,
      "loss": 0.4392,
      "step": 957
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2025893926620483,
      "learning_rate": 9.66380707472348e-06,
      "loss": 0.5102,
      "step": 958
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.070096731185913,
      "learning_rate": 9.662929547172575e-06,
      "loss": 0.4692,
      "step": 959
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4417049884796143,
      "learning_rate": 9.662050915804826e-06,
      "loss": 0.5536,
      "step": 960
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.8226470947265625,
      "learning_rate": 9.661171180828221e-06,
      "loss": 0.4945,
      "step": 961
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.087707281112671,
      "learning_rate": 9.660290342451015e-06,
      "loss": 0.4803,
      "step": 962
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.038534164428711,
      "learning_rate": 9.659408400881722e-06,
      "loss": 0.4272,
      "step": 963
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5749986171722412,
      "learning_rate": 9.658525356329113e-06,
      "loss": 0.4454,
      "step": 964
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9816405177116394,
      "learning_rate": 9.657641209002226e-06,
      "loss": 0.5217,
      "step": 965
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.304269790649414,
      "learning_rate": 9.656755959110359e-06,
      "loss": 0.4481,
      "step": 966
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1544514894485474,
      "learning_rate": 9.655869606863068e-06,
      "loss": 0.5118,
      "step": 967
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4427409172058105,
      "learning_rate": 9.654982152470173e-06,
      "loss": 0.4407,
      "step": 968
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.1910667419433594,
      "learning_rate": 9.654093596141753e-06,
      "loss": 0.4646,
      "step": 969
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.663752794265747,
      "learning_rate": 9.65320393808815e-06,
      "loss": 0.4665,
      "step": 970
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3470394611358643,
      "learning_rate": 9.652313178519962e-06,
      "loss": 0.5184,
      "step": 971
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1871132850646973,
      "learning_rate": 9.651421317648058e-06,
      "loss": 0.4077,
      "step": 972
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3750030994415283,
      "learning_rate": 9.650528355683553e-06,
      "loss": 0.5097,
      "step": 973
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7359870672225952,
      "learning_rate": 9.649634292837834e-06,
      "loss": 0.5075,
      "step": 974
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.019329309463501,
      "learning_rate": 9.648739129322547e-06,
      "loss": 0.4842,
      "step": 975
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5896633863449097,
      "learning_rate": 9.647842865349595e-06,
      "loss": 0.5539,
      "step": 976
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4290679693222046,
      "learning_rate": 9.646945501131142e-06,
      "loss": 0.467,
      "step": 977
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1497743129730225,
      "learning_rate": 9.646047036879614e-06,
      "loss": 0.5541,
      "step": 978
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2641074657440186,
      "learning_rate": 9.645147472807698e-06,
      "loss": 0.5122,
      "step": 979
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1433510780334473,
      "learning_rate": 9.644246809128342e-06,
      "loss": 0.4778,
      "step": 980
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.480405569076538,
      "learning_rate": 9.643345046054748e-06,
      "loss": 0.5871,
      "step": 981
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7255730628967285,
      "learning_rate": 9.642442183800387e-06,
      "loss": 0.4909,
      "step": 982
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4446533918380737,
      "learning_rate": 9.641538222578984e-06,
      "loss": 0.5159,
      "step": 983
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1474405527114868,
      "learning_rate": 9.640633162604528e-06,
      "loss": 0.4284,
      "step": 984
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1062803268432617,
      "learning_rate": 9.639727004091265e-06,
      "loss": 0.4504,
      "step": 985
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8532858490943909,
      "learning_rate": 9.638819747253703e-06,
      "loss": 0.3982,
      "step": 986
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1064850091934204,
      "learning_rate": 9.63791139230661e-06,
      "loss": 0.4826,
      "step": 987
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8917860984802246,
      "learning_rate": 9.637001939465012e-06,
      "loss": 0.4597,
      "step": 988
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.079199194908142,
      "learning_rate": 9.636091388944195e-06,
      "loss": 0.4451,
      "step": 989
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.085074543952942,
      "learning_rate": 9.63517974095971e-06,
      "loss": 0.5071,
      "step": 990
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5940099954605103,
      "learning_rate": 9.634266995727364e-06,
      "loss": 0.5481,
      "step": 991
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6415232419967651,
      "learning_rate": 9.633353153463218e-06,
      "loss": 0.4727,
      "step": 992
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5828561782836914,
      "learning_rate": 9.632438214383604e-06,
      "loss": 0.6043,
      "step": 993
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.373353123664856,
      "learning_rate": 9.631522178705106e-06,
      "loss": 0.4786,
      "step": 994
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3473150730133057,
      "learning_rate": 9.630605046644569e-06,
      "loss": 0.4843,
      "step": 995
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3530644178390503,
      "learning_rate": 9.629686818419099e-06,
      "loss": 0.5091,
      "step": 996
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1787447929382324,
      "learning_rate": 9.628767494246059e-06,
      "loss": 0.4625,
      "step": 997
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1521248817443848,
      "learning_rate": 9.627847074343075e-06,
      "loss": 0.4838,
      "step": 998
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.161715030670166,
      "learning_rate": 9.62692555892803e-06,
      "loss": 0.4846,
      "step": 999
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7483254671096802,
      "learning_rate": 9.626002948219067e-06,
      "loss": 0.4855,
      "step": 1000
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4208906888961792,
      "learning_rate": 9.625079242434585e-06,
      "loss": 0.528,
      "step": 1001
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.720930576324463,
      "learning_rate": 9.62415444179325e-06,
      "loss": 0.4638,
      "step": 1002
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.254412293434143,
      "learning_rate": 9.623228546513978e-06,
      "loss": 0.5396,
      "step": 1003
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2825509309768677,
      "learning_rate": 9.622301556815953e-06,
      "loss": 0.4548,
      "step": 1004
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5098316669464111,
      "learning_rate": 9.62137347291861e-06,
      "loss": 0.454,
      "step": 1005
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3123667240142822,
      "learning_rate": 9.620444295041649e-06,
      "loss": 0.5324,
      "step": 1006
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0257368087768555,
      "learning_rate": 9.619514023405023e-06,
      "loss": 0.4526,
      "step": 1007
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2446401119232178,
      "learning_rate": 9.618582658228951e-06,
      "loss": 0.4972,
      "step": 1008
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.126914143562317,
      "learning_rate": 9.617650199733909e-06,
      "loss": 0.5487,
      "step": 1009
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1372220516204834,
      "learning_rate": 9.616716648140625e-06,
      "loss": 0.4572,
      "step": 1010
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1262078285217285,
      "learning_rate": 9.615782003670095e-06,
      "loss": 0.4912,
      "step": 1011
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0723237991333008,
      "learning_rate": 9.614846266543568e-06,
      "loss": 0.5047,
      "step": 1012
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3325002193450928,
      "learning_rate": 9.613909436982553e-06,
      "loss": 0.72,
      "step": 1013
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2758649587631226,
      "learning_rate": 9.61297151520882e-06,
      "loss": 0.4502,
      "step": 1014
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1754767894744873,
      "learning_rate": 9.612032501444392e-06,
      "loss": 0.4672,
      "step": 1015
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1717678308486938,
      "learning_rate": 9.611092395911556e-06,
      "loss": 0.4272,
      "step": 1016
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0513219833374023,
      "learning_rate": 9.610151198832857e-06,
      "loss": 0.4801,
      "step": 1017
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4781384468078613,
      "learning_rate": 9.609208910431096e-06,
      "loss": 0.5408,
      "step": 1018
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6400364637374878,
      "learning_rate": 9.60826553092933e-06,
      "loss": 0.4794,
      "step": 1019
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2921844720840454,
      "learning_rate": 9.607321060550881e-06,
      "loss": 0.4419,
      "step": 1020
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5674574375152588,
      "learning_rate": 9.606375499519325e-06,
      "loss": 0.4533,
      "step": 1021
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2642717361450195,
      "learning_rate": 9.605428848058497e-06,
      "loss": 0.5121,
      "step": 1022
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.9269797801971436,
      "learning_rate": 9.604481106392489e-06,
      "loss": 0.4625,
      "step": 1023
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.320719838142395,
      "learning_rate": 9.603532274745651e-06,
      "loss": 0.5203,
      "step": 1024
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1357622146606445,
      "learning_rate": 9.602582353342595e-06,
      "loss": 0.4648,
      "step": 1025
    },
    {
      "epoch": 0.15,
      "grad_norm": 12.395257949829102,
      "learning_rate": 9.601631342408186e-06,
      "loss": 0.5644,
      "step": 1026
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.373812198638916,
      "learning_rate": 9.60067924216755e-06,
      "loss": 0.4695,
      "step": 1027
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5407283306121826,
      "learning_rate": 9.599726052846069e-06,
      "loss": 0.4915,
      "step": 1028
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.150713324546814,
      "learning_rate": 9.598771774669384e-06,
      "loss": 0.5181,
      "step": 1029
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.284599542617798,
      "learning_rate": 9.597816407863395e-06,
      "loss": 0.5209,
      "step": 1030
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4366945028305054,
      "learning_rate": 9.596859952654255e-06,
      "loss": 0.4735,
      "step": 1031
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3741298913955688,
      "learning_rate": 9.59590240926838e-06,
      "loss": 0.4645,
      "step": 1032
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1706948280334473,
      "learning_rate": 9.594943777932441e-06,
      "loss": 0.5137,
      "step": 1033
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5650242567062378,
      "learning_rate": 9.593984058873365e-06,
      "loss": 0.5318,
      "step": 1034
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2364846467971802,
      "learning_rate": 9.59302325231834e-06,
      "loss": 0.4942,
      "step": 1035
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.266636610031128,
      "learning_rate": 9.592061358494814e-06,
      "loss": 0.5475,
      "step": 1036
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3188433647155762,
      "learning_rate": 9.59109837763048e-06,
      "loss": 0.7522,
      "step": 1037
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5629489421844482,
      "learning_rate": 9.5901343099533e-06,
      "loss": 0.4993,
      "step": 1038
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7561697959899902,
      "learning_rate": 9.589169155691491e-06,
      "loss": 0.5412,
      "step": 1039
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.215700387954712,
      "learning_rate": 9.588202915073526e-06,
      "loss": 0.4404,
      "step": 1040
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0306310653686523,
      "learning_rate": 9.587235588328134e-06,
      "loss": 0.5106,
      "step": 1041
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1843163967132568,
      "learning_rate": 9.586267175684303e-06,
      "loss": 0.492,
      "step": 1042
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3044947385787964,
      "learning_rate": 9.585297677371277e-06,
      "loss": 0.5021,
      "step": 1043
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3993301391601562,
      "learning_rate": 9.584327093618558e-06,
      "loss": 0.5003,
      "step": 1044
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.171261191368103,
      "learning_rate": 9.583355424655903e-06,
      "loss": 0.5404,
      "step": 1045
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6644823551177979,
      "learning_rate": 9.582382670713328e-06,
      "loss": 0.4849,
      "step": 1046
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3558210134506226,
      "learning_rate": 9.581408832021106e-06,
      "loss": 0.5378,
      "step": 1047
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.455425500869751,
      "learning_rate": 9.580433908809763e-06,
      "loss": 0.4541,
      "step": 1048
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1595910787582397,
      "learning_rate": 9.579457901310086e-06,
      "loss": 0.4538,
      "step": 1049
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1499059200286865,
      "learning_rate": 9.578480809753118e-06,
      "loss": 0.443,
      "step": 1050
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.154876947402954,
      "learning_rate": 9.577502634370157e-06,
      "loss": 0.7493,
      "step": 1051
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.564640760421753,
      "learning_rate": 9.57652337539276e-06,
      "loss": 0.5467,
      "step": 1052
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2102971076965332,
      "learning_rate": 9.575543033052736e-06,
      "loss": 0.4591,
      "step": 1053
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2469761371612549,
      "learning_rate": 9.574561607582156e-06,
      "loss": 0.7345,
      "step": 1054
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2472519874572754,
      "learning_rate": 9.573579099213342e-06,
      "loss": 0.4991,
      "step": 1055
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.10712730884552,
      "learning_rate": 9.572595508178877e-06,
      "loss": 0.5521,
      "step": 1056
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.599442720413208,
      "learning_rate": 9.5716108347116e-06,
      "loss": 0.5356,
      "step": 1057
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3637969493865967,
      "learning_rate": 9.570625079044601e-06,
      "loss": 0.4905,
      "step": 1058
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.110664963722229,
      "learning_rate": 9.569638241411232e-06,
      "loss": 0.4631,
      "step": 1059
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1579445600509644,
      "learning_rate": 9.568650322045098e-06,
      "loss": 0.4773,
      "step": 1060
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.198945164680481,
      "learning_rate": 9.567661321180064e-06,
      "loss": 0.4704,
      "step": 1061
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5837699174880981,
      "learning_rate": 9.566671239050243e-06,
      "loss": 0.5052,
      "step": 1062
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.085720181465149,
      "learning_rate": 9.565680075890011e-06,
      "loss": 0.6996,
      "step": 1063
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1354708671569824,
      "learning_rate": 9.564687831934e-06,
      "loss": 0.4911,
      "step": 1064
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8081507682800293,
      "learning_rate": 9.563694507417094e-06,
      "loss": 0.5372,
      "step": 1065
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.417773723602295,
      "learning_rate": 9.562700102574434e-06,
      "loss": 0.459,
      "step": 1066
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.436977744102478,
      "learning_rate": 9.561704617641417e-06,
      "loss": 0.4718,
      "step": 1067
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.115619421005249,
      "learning_rate": 9.560708052853697e-06,
      "loss": 0.5305,
      "step": 1068
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4509015083312988,
      "learning_rate": 9.559710408447185e-06,
      "loss": 0.5985,
      "step": 1069
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.233585000038147,
      "learning_rate": 9.55871168465804e-06,
      "loss": 0.4812,
      "step": 1070
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1908599138259888,
      "learning_rate": 9.557711881722684e-06,
      "loss": 0.4963,
      "step": 1071
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3039112091064453,
      "learning_rate": 9.556710999877795e-06,
      "loss": 0.4681,
      "step": 1072
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0899412631988525,
      "learning_rate": 9.555709039360297e-06,
      "loss": 0.4781,
      "step": 1073
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3949179649353027,
      "learning_rate": 9.554706000407382e-06,
      "loss": 0.5124,
      "step": 1074
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1844686269760132,
      "learning_rate": 9.553701883256488e-06,
      "loss": 0.5341,
      "step": 1075
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.273400902748108,
      "learning_rate": 9.552696688145313e-06,
      "loss": 0.5327,
      "step": 1076
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5647404193878174,
      "learning_rate": 9.551690415311806e-06,
      "loss": 0.5265,
      "step": 1077
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3262625932693481,
      "learning_rate": 9.550683064994176e-06,
      "loss": 0.4981,
      "step": 1078
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2116751670837402,
      "learning_rate": 9.549674637430884e-06,
      "loss": 0.5461,
      "step": 1079
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1552929878234863,
      "learning_rate": 9.548665132860648e-06,
      "loss": 0.5459,
      "step": 1080
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.728712558746338,
      "learning_rate": 9.547654551522436e-06,
      "loss": 0.5131,
      "step": 1081
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.349717617034912,
      "learning_rate": 9.54664289365548e-06,
      "loss": 0.475,
      "step": 1082
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.367572546005249,
      "learning_rate": 9.545630159499254e-06,
      "loss": 0.5039,
      "step": 1083
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6675405502319336,
      "learning_rate": 9.544616349293503e-06,
      "loss": 0.455,
      "step": 1084
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0868117809295654,
      "learning_rate": 9.543601463278212e-06,
      "loss": 0.4853,
      "step": 1085
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3948006629943848,
      "learning_rate": 9.542585501693628e-06,
      "loss": 0.5478,
      "step": 1086
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1446213722229004,
      "learning_rate": 9.541568464780252e-06,
      "loss": 0.5012,
      "step": 1087
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1785314083099365,
      "learning_rate": 9.540550352778838e-06,
      "loss": 0.5151,
      "step": 1088
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6304492950439453,
      "learning_rate": 9.539531165930396e-06,
      "loss": 0.5379,
      "step": 1089
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4124155044555664,
      "learning_rate": 9.538510904476188e-06,
      "loss": 0.5122,
      "step": 1090
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2260017395019531,
      "learning_rate": 9.537489568657734e-06,
      "loss": 0.5184,
      "step": 1091
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.099176049232483,
      "learning_rate": 9.536467158716804e-06,
      "loss": 0.4442,
      "step": 1092
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2905299663543701,
      "learning_rate": 9.535443674895428e-06,
      "loss": 0.4432,
      "step": 1093
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1579813957214355,
      "learning_rate": 9.534419117435885e-06,
      "loss": 0.4644,
      "step": 1094
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4355016946792603,
      "learning_rate": 9.53339348658071e-06,
      "loss": 0.4903,
      "step": 1095
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3738434314727783,
      "learning_rate": 9.53236678257269e-06,
      "loss": 0.5164,
      "step": 1096
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2521355152130127,
      "learning_rate": 9.531339005654873e-06,
      "loss": 0.4548,
      "step": 1097
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4429285526275635,
      "learning_rate": 9.530310156070554e-06,
      "loss": 0.4223,
      "step": 1098
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1352076530456543,
      "learning_rate": 9.529280234063283e-06,
      "loss": 0.4231,
      "step": 1099
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.227839469909668,
      "learning_rate": 9.528249239876866e-06,
      "loss": 0.5005,
      "step": 1100
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3958613872528076,
      "learning_rate": 9.52721717375536e-06,
      "loss": 0.5249,
      "step": 1101
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.315597653388977,
      "learning_rate": 9.52618403594308e-06,
      "loss": 0.7896,
      "step": 1102
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1102919578552246,
      "learning_rate": 9.52514982668459e-06,
      "loss": 0.4003,
      "step": 1103
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2847667932510376,
      "learning_rate": 9.524114546224713e-06,
      "loss": 0.5327,
      "step": 1104
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3386354446411133,
      "learning_rate": 9.523078194808518e-06,
      "loss": 0.4863,
      "step": 1105
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2271502017974854,
      "learning_rate": 9.522040772681336e-06,
      "loss": 0.4825,
      "step": 1106
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.418238639831543,
      "learning_rate": 9.521002280088745e-06,
      "loss": 0.4599,
      "step": 1107
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4414945840835571,
      "learning_rate": 9.519962717276578e-06,
      "loss": 0.4956,
      "step": 1108
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5429145097732544,
      "learning_rate": 9.518922084490925e-06,
      "loss": 0.5223,
      "step": 1109
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.036512017250061,
      "learning_rate": 9.517880381978125e-06,
      "loss": 0.5372,
      "step": 1110
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1291637420654297,
      "learning_rate": 9.516837609984771e-06,
      "loss": 0.4184,
      "step": 1111
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.183877944946289,
      "learning_rate": 9.515793768757711e-06,
      "loss": 0.4807,
      "step": 1112
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4311884641647339,
      "learning_rate": 9.514748858544044e-06,
      "loss": 0.4826,
      "step": 1113
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1258922815322876,
      "learning_rate": 9.513702879591124e-06,
      "loss": 0.4616,
      "step": 1114
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1468640565872192,
      "learning_rate": 9.512655832146556e-06,
      "loss": 0.4708,
      "step": 1115
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4161655902862549,
      "learning_rate": 9.511607716458198e-06,
      "loss": 0.4397,
      "step": 1116
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.138113260269165,
      "learning_rate": 9.510558532774163e-06,
      "loss": 0.5046,
      "step": 1117
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.526832103729248,
      "learning_rate": 9.509508281342816e-06,
      "loss": 0.4783,
      "step": 1118
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4346390962600708,
      "learning_rate": 9.508456962412775e-06,
      "loss": 0.5078,
      "step": 1119
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4143728017807007,
      "learning_rate": 9.507404576232908e-06,
      "loss": 0.4834,
      "step": 1120
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.347864031791687,
      "learning_rate": 9.506351123052339e-06,
      "loss": 0.7925,
      "step": 1121
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5606662034988403,
      "learning_rate": 9.505296603120443e-06,
      "loss": 0.4426,
      "step": 1122
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2195130586624146,
      "learning_rate": 9.504241016686847e-06,
      "loss": 0.5314,
      "step": 1123
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.20423424243927,
      "learning_rate": 9.503184364001432e-06,
      "loss": 0.5331,
      "step": 1124
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.048647880554199,
      "learning_rate": 9.50212664531433e-06,
      "loss": 0.5154,
      "step": 1125
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0093634128570557,
      "learning_rate": 9.50106786087593e-06,
      "loss": 0.4354,
      "step": 1126
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3409005403518677,
      "learning_rate": 9.500008010936863e-06,
      "loss": 0.5574,
      "step": 1127
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6969184875488281,
      "learning_rate": 9.498947095748023e-06,
      "loss": 0.4779,
      "step": 1128
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2931787967681885,
      "learning_rate": 9.49788511556055e-06,
      "loss": 0.492,
      "step": 1129
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1753178834915161,
      "learning_rate": 9.496822070625838e-06,
      "loss": 0.4847,
      "step": 1130
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.112696886062622,
      "learning_rate": 9.495757961195533e-06,
      "loss": 0.4599,
      "step": 1131
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.15277099609375,
      "learning_rate": 9.494692787521531e-06,
      "loss": 0.515,
      "step": 1132
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2694482803344727,
      "learning_rate": 9.493626549855986e-06,
      "loss": 0.4869,
      "step": 1133
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.016261100769043,
      "learning_rate": 9.492559248451297e-06,
      "loss": 0.4664,
      "step": 1134
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2892043590545654,
      "learning_rate": 9.491490883560116e-06,
      "loss": 0.4728,
      "step": 1135
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3270938396453857,
      "learning_rate": 9.49042145543535e-06,
      "loss": 0.5419,
      "step": 1136
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.173202633857727,
      "learning_rate": 9.489350964330157e-06,
      "loss": 0.7265,
      "step": 1137
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.174923300743103,
      "learning_rate": 9.488279410497944e-06,
      "loss": 0.5293,
      "step": 1138
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.199428915977478,
      "learning_rate": 9.48720679419237e-06,
      "loss": 0.5445,
      "step": 1139
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.345248818397522,
      "learning_rate": 9.486133115667349e-06,
      "loss": 0.461,
      "step": 1140
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2446491718292236,
      "learning_rate": 9.485058375177044e-06,
      "loss": 0.4445,
      "step": 1141
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1231244802474976,
      "learning_rate": 9.483982572975868e-06,
      "loss": 0.4678,
      "step": 1142
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2045564651489258,
      "learning_rate": 9.482905709318486e-06,
      "loss": 0.47,
      "step": 1143
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3197152614593506,
      "learning_rate": 9.481827784459818e-06,
      "loss": 0.4555,
      "step": 1144
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2856472730636597,
      "learning_rate": 9.48074879865503e-06,
      "loss": 0.7573,
      "step": 1145
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.685503363609314,
      "learning_rate": 9.47966875215954e-06,
      "loss": 0.5557,
      "step": 1146
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1354399919509888,
      "learning_rate": 9.478587645229023e-06,
      "loss": 0.4281,
      "step": 1147
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3365647792816162,
      "learning_rate": 9.477505478119398e-06,
      "loss": 0.5219,
      "step": 1148
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.246524691581726,
      "learning_rate": 9.476422251086837e-06,
      "loss": 0.4189,
      "step": 1149
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1277300119400024,
      "learning_rate": 9.475337964387763e-06,
      "loss": 0.4224,
      "step": 1150
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4024416208267212,
      "learning_rate": 9.474252618278852e-06,
      "loss": 0.5296,
      "step": 1151
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2442841529846191,
      "learning_rate": 9.473166213017029e-06,
      "loss": 0.7225,
      "step": 1152
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.151771903038025,
      "learning_rate": 9.472078748859467e-06,
      "loss": 0.4863,
      "step": 1153
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4284719228744507,
      "learning_rate": 9.470990226063596e-06,
      "loss": 0.5512,
      "step": 1154
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5212942361831665,
      "learning_rate": 9.469900644887092e-06,
      "loss": 0.468,
      "step": 1155
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.347912073135376,
      "learning_rate": 9.468810005587879e-06,
      "loss": 0.7817,
      "step": 1156
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7097488641738892,
      "learning_rate": 9.467718308424139e-06,
      "loss": 0.4853,
      "step": 1157
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3495978116989136,
      "learning_rate": 9.4666255536543e-06,
      "loss": 0.4649,
      "step": 1158
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0819438695907593,
      "learning_rate": 9.465531741537041e-06,
      "loss": 0.4714,
      "step": 1159
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3234589099884033,
      "learning_rate": 9.464436872331291e-06,
      "loss": 0.4539,
      "step": 1160
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0249155759811401,
      "learning_rate": 9.463340946296228e-06,
      "loss": 0.4358,
      "step": 1161
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1415079832077026,
      "learning_rate": 9.46224396369128e-06,
      "loss": 0.4683,
      "step": 1162
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.33318293094635,
      "learning_rate": 9.461145924776131e-06,
      "loss": 0.5012,
      "step": 1163
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3253921270370483,
      "learning_rate": 9.46004682981071e-06,
      "loss": 0.5301,
      "step": 1164
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1902514696121216,
      "learning_rate": 9.458946679055192e-06,
      "loss": 0.4699,
      "step": 1165
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.663066029548645,
      "learning_rate": 9.457845472770012e-06,
      "loss": 0.5004,
      "step": 1166
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1824227571487427,
      "learning_rate": 9.456743211215846e-06,
      "loss": 0.5056,
      "step": 1167
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.545220971107483,
      "learning_rate": 9.455639894653627e-06,
      "loss": 0.5388,
      "step": 1168
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9165988564491272,
      "learning_rate": 9.45453552334453e-06,
      "loss": 0.4832,
      "step": 1169
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1253265142440796,
      "learning_rate": 9.453430097549986e-06,
      "loss": 0.4415,
      "step": 1170
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1628572940826416,
      "learning_rate": 9.452323617531673e-06,
      "loss": 0.7547,
      "step": 1171
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0012823343276978,
      "learning_rate": 9.451216083551517e-06,
      "loss": 0.4444,
      "step": 1172
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6043733358383179,
      "learning_rate": 9.450107495871699e-06,
      "loss": 0.4691,
      "step": 1173
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.458067536354065,
      "learning_rate": 9.448997854754643e-06,
      "loss": 0.4597,
      "step": 1174
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2056562900543213,
      "learning_rate": 9.447887160463025e-06,
      "loss": 0.7286,
      "step": 1175
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1714786291122437,
      "learning_rate": 9.446775413259773e-06,
      "loss": 0.476,
      "step": 1176
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6545659303665161,
      "learning_rate": 9.44566261340806e-06,
      "loss": 0.4989,
      "step": 1177
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4042901992797852,
      "learning_rate": 9.44454876117131e-06,
      "loss": 0.4705,
      "step": 1178
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.21773099899292,
      "learning_rate": 9.443433856813197e-06,
      "loss": 0.5194,
      "step": 1179
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1964210271835327,
      "learning_rate": 9.442317900597641e-06,
      "loss": 0.4563,
      "step": 1180
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1883326768875122,
      "learning_rate": 9.441200892788815e-06,
      "loss": 0.4828,
      "step": 1181
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.932890772819519,
      "learning_rate": 9.440082833651138e-06,
      "loss": 0.4885,
      "step": 1182
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3924930095672607,
      "learning_rate": 9.438963723449281e-06,
      "loss": 0.4227,
      "step": 1183
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3564198017120361,
      "learning_rate": 9.437843562448158e-06,
      "loss": 0.4068,
      "step": 1184
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9835208654403687,
      "learning_rate": 9.43672235091294e-06,
      "loss": 0.4503,
      "step": 1185
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2485971450805664,
      "learning_rate": 9.435600089109038e-06,
      "loss": 0.4851,
      "step": 1186
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3195878267288208,
      "learning_rate": 9.434476777302119e-06,
      "loss": 0.4562,
      "step": 1187
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.014068365097046,
      "learning_rate": 9.433352415758093e-06,
      "loss": 0.5293,
      "step": 1188
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3335466384887695,
      "learning_rate": 9.432227004743122e-06,
      "loss": 0.4407,
      "step": 1189
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2301685810089111,
      "learning_rate": 9.431100544523614e-06,
      "loss": 0.5129,
      "step": 1190
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0924451351165771,
      "learning_rate": 9.429973035366228e-06,
      "loss": 0.7173,
      "step": 1191
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.499753475189209,
      "learning_rate": 9.42884447753787e-06,
      "loss": 0.4762,
      "step": 1192
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1971774101257324,
      "learning_rate": 9.427714871305694e-06,
      "loss": 0.434,
      "step": 1193
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2081612348556519,
      "learning_rate": 9.426584216937103e-06,
      "loss": 0.5176,
      "step": 1194
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.324704647064209,
      "learning_rate": 9.425452514699745e-06,
      "loss": 0.5013,
      "step": 1195
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5192703008651733,
      "learning_rate": 9.42431976486152e-06,
      "loss": 0.5193,
      "step": 1196
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.628275990486145,
      "learning_rate": 9.423185967690575e-06,
      "loss": 0.4856,
      "step": 1197
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3709508180618286,
      "learning_rate": 9.422051123455305e-06,
      "loss": 0.4907,
      "step": 1198
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.498291015625,
      "learning_rate": 9.420915232424351e-06,
      "loss": 0.4827,
      "step": 1199
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6805729866027832,
      "learning_rate": 9.419778294866603e-06,
      "loss": 0.4189,
      "step": 1200
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4906257390975952,
      "learning_rate": 9.418640311051199e-06,
      "loss": 0.5351,
      "step": 1201
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2063584327697754,
      "learning_rate": 9.417501281247526e-06,
      "loss": 0.5438,
      "step": 1202
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3997817039489746,
      "learning_rate": 9.416361205725216e-06,
      "loss": 0.4992,
      "step": 1203
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2815519571304321,
      "learning_rate": 9.41522008475415e-06,
      "loss": 0.5025,
      "step": 1204
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1126173734664917,
      "learning_rate": 9.414077918604455e-06,
      "loss": 0.4874,
      "step": 1205
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.353308081626892,
      "learning_rate": 9.412934707546508e-06,
      "loss": 0.4035,
      "step": 1206
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.306027889251709,
      "learning_rate": 9.411790451850931e-06,
      "loss": 0.4533,
      "step": 1207
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1209601163864136,
      "learning_rate": 9.410645151788595e-06,
      "loss": 0.3951,
      "step": 1208
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.290703535079956,
      "learning_rate": 9.409498807630616e-06,
      "loss": 0.4755,
      "step": 1209
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6387453079223633,
      "learning_rate": 9.408351419648361e-06,
      "loss": 0.4965,
      "step": 1210
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6973243951797485,
      "learning_rate": 9.40720298811344e-06,
      "loss": 0.5208,
      "step": 1211
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6782242059707642,
      "learning_rate": 9.40605351329771e-06,
      "loss": 0.4632,
      "step": 1212
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5205553770065308,
      "learning_rate": 9.404902995473282e-06,
      "loss": 0.46,
      "step": 1213
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9928238987922668,
      "learning_rate": 9.403751434912503e-06,
      "loss": 0.466,
      "step": 1214
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.4854516983032227,
      "learning_rate": 9.402598831887974e-06,
      "loss": 0.5147,
      "step": 1215
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6045159101486206,
      "learning_rate": 9.401445186672544e-06,
      "loss": 0.4674,
      "step": 1216
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2390836477279663,
      "learning_rate": 9.400290499539303e-06,
      "loss": 0.515,
      "step": 1217
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4382315874099731,
      "learning_rate": 9.399134770761592e-06,
      "loss": 0.5377,
      "step": 1218
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1563266515731812,
      "learning_rate": 9.397978000612998e-06,
      "loss": 0.4732,
      "step": 1219
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3144603967666626,
      "learning_rate": 9.396820189367349e-06,
      "loss": 0.5312,
      "step": 1220
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.224370002746582,
      "learning_rate": 9.395661337298729e-06,
      "loss": 0.4544,
      "step": 1221
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.451256275177002,
      "learning_rate": 9.394501444681461e-06,
      "loss": 0.4225,
      "step": 1222
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0506609678268433,
      "learning_rate": 9.393340511790117e-06,
      "loss": 0.4471,
      "step": 1223
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.570957064628601,
      "learning_rate": 9.392178538899518e-06,
      "loss": 0.4881,
      "step": 1224
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0684528350830078,
      "learning_rate": 9.391015526284724e-06,
      "loss": 0.4348,
      "step": 1225
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5863797664642334,
      "learning_rate": 9.389851474221047e-06,
      "loss": 0.5097,
      "step": 1226
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2880173921585083,
      "learning_rate": 9.388686382984047e-06,
      "loss": 0.5084,
      "step": 1227
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1452208757400513,
      "learning_rate": 9.38752025284952e-06,
      "loss": 0.5077,
      "step": 1228
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6081522703170776,
      "learning_rate": 9.38635308409352e-06,
      "loss": 0.504,
      "step": 1229
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3270848989486694,
      "learning_rate": 9.385184876992337e-06,
      "loss": 0.5386,
      "step": 1230
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2466723918914795,
      "learning_rate": 9.384015631822512e-06,
      "loss": 0.7377,
      "step": 1231
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.268829584121704,
      "learning_rate": 9.382845348860834e-06,
      "loss": 0.4938,
      "step": 1232
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3288289308547974,
      "learning_rate": 9.38167402838433e-06,
      "loss": 0.499,
      "step": 1233
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1426506042480469,
      "learning_rate": 9.38050167067028e-06,
      "loss": 0.465,
      "step": 1234
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2496495246887207,
      "learning_rate": 9.379328275996206e-06,
      "loss": 0.4652,
      "step": 1235
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2920503616333008,
      "learning_rate": 9.378153844639875e-06,
      "loss": 0.4722,
      "step": 1236
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2921335697174072,
      "learning_rate": 9.376978376879302e-06,
      "loss": 0.4761,
      "step": 1237
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2928822040557861,
      "learning_rate": 9.375801872992744e-06,
      "loss": 0.4678,
      "step": 1238
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1694051027297974,
      "learning_rate": 9.374624333258707e-06,
      "loss": 0.4566,
      "step": 1239
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1053593158721924,
      "learning_rate": 9.373445757955937e-06,
      "loss": 0.5061,
      "step": 1240
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0264523029327393,
      "learning_rate": 9.372266147363433e-06,
      "loss": 0.4619,
      "step": 1241
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.214593768119812,
      "learning_rate": 9.37108550176043e-06,
      "loss": 0.4591,
      "step": 1242
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1678489446640015,
      "learning_rate": 9.369903821426416e-06,
      "loss": 0.4932,
      "step": 1243
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1252334117889404,
      "learning_rate": 9.368721106641119e-06,
      "loss": 0.5177,
      "step": 1244
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1536844968795776,
      "learning_rate": 9.367537357684515e-06,
      "loss": 0.4819,
      "step": 1245
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1767069101333618,
      "learning_rate": 9.366352574836819e-06,
      "loss": 0.4588,
      "step": 1246
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1060625314712524,
      "learning_rate": 9.3651667583785e-06,
      "loss": 0.7101,
      "step": 1247
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2360365390777588,
      "learning_rate": 9.36397990859026e-06,
      "loss": 0.7722,
      "step": 1248
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1012039184570312,
      "learning_rate": 9.362792025753059e-06,
      "loss": 0.5318,
      "step": 1249
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5950000286102295,
      "learning_rate": 9.36160311014809e-06,
      "loss": 0.5347,
      "step": 1250
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1903048753738403,
      "learning_rate": 9.360413162056798e-06,
      "loss": 0.4212,
      "step": 1251
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0074949264526367,
      "learning_rate": 9.359222181760869e-06,
      "loss": 0.4754,
      "step": 1252
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1472924947738647,
      "learning_rate": 9.358030169542235e-06,
      "loss": 0.4768,
      "step": 1253
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1856054067611694,
      "learning_rate": 9.356837125683065e-06,
      "loss": 0.4644,
      "step": 1254
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3158994913101196,
      "learning_rate": 9.355643050465786e-06,
      "loss": 0.3963,
      "step": 1255
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1216861009597778,
      "learning_rate": 9.354447944173059e-06,
      "loss": 0.5393,
      "step": 1256
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0125043392181396,
      "learning_rate": 9.353251807087792e-06,
      "loss": 0.5335,
      "step": 1257
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0903682708740234,
      "learning_rate": 9.352054639493133e-06,
      "loss": 0.5205,
      "step": 1258
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2000842094421387,
      "learning_rate": 9.350856441672484e-06,
      "loss": 0.4927,
      "step": 1259
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1490895748138428,
      "learning_rate": 9.34965721390948e-06,
      "loss": 0.5098,
      "step": 1260
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2238926887512207,
      "learning_rate": 9.348456956488005e-06,
      "loss": 0.4715,
      "step": 1261
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0808016061782837,
      "learning_rate": 9.347255669692186e-06,
      "loss": 0.4495,
      "step": 1262
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5139726400375366,
      "learning_rate": 9.346053353806394e-06,
      "loss": 0.574,
      "step": 1263
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3480772972106934,
      "learning_rate": 9.344850009115244e-06,
      "loss": 0.4706,
      "step": 1264
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.264905333518982,
      "learning_rate": 9.343645635903595e-06,
      "loss": 0.4918,
      "step": 1265
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5346323251724243,
      "learning_rate": 9.342440234456546e-06,
      "loss": 0.4967,
      "step": 1266
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.280221700668335,
      "learning_rate": 9.341233805059442e-06,
      "loss": 0.439,
      "step": 1267
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6719764471054077,
      "learning_rate": 9.340026347997872e-06,
      "loss": 0.4547,
      "step": 1268
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3922406435012817,
      "learning_rate": 9.338817863557665e-06,
      "loss": 0.4231,
      "step": 1269
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4782390594482422,
      "learning_rate": 9.3376083520249e-06,
      "loss": 0.4869,
      "step": 1270
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1608601808547974,
      "learning_rate": 9.336397813685891e-06,
      "loss": 0.4966,
      "step": 1271
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1291979551315308,
      "learning_rate": 9.335186248827199e-06,
      "loss": 0.4998,
      "step": 1272
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.042586088180542,
      "learning_rate": 9.333973657735629e-06,
      "loss": 0.5,
      "step": 1273
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2445436716079712,
      "learning_rate": 9.332760040698228e-06,
      "loss": 0.4776,
      "step": 1274
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0780541896820068,
      "learning_rate": 9.331545398002285e-06,
      "loss": 0.521,
      "step": 1275
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3488092422485352,
      "learning_rate": 9.330329729935332e-06,
      "loss": 0.5046,
      "step": 1276
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2317440509796143,
      "learning_rate": 9.329113036785144e-06,
      "loss": 0.4909,
      "step": 1277
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.131848931312561,
      "learning_rate": 9.32789531883974e-06,
      "loss": 0.5037,
      "step": 1278
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1176602840423584,
      "learning_rate": 9.326676576387377e-06,
      "loss": 0.461,
      "step": 1279
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.108501672744751,
      "learning_rate": 9.325456809716562e-06,
      "loss": 0.4655,
      "step": 1280
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6823890209197998,
      "learning_rate": 9.324236019116038e-06,
      "loss": 0.5296,
      "step": 1281
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4474070072174072,
      "learning_rate": 9.323014204874793e-06,
      "loss": 0.4898,
      "step": 1282
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.64805269241333,
      "learning_rate": 9.321791367282058e-06,
      "loss": 0.5327,
      "step": 1283
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2114120721817017,
      "learning_rate": 9.320567506627302e-06,
      "loss": 0.7304,
      "step": 1284
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7735166549682617,
      "learning_rate": 9.319342623200245e-06,
      "loss": 0.4856,
      "step": 1285
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.333343505859375,
      "learning_rate": 9.318116717290839e-06,
      "loss": 0.5057,
      "step": 1286
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.244326114654541,
      "learning_rate": 9.316889789189285e-06,
      "loss": 0.5094,
      "step": 1287
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2931450605392456,
      "learning_rate": 9.315661839186022e-06,
      "loss": 0.5309,
      "step": 1288
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5515788793563843,
      "learning_rate": 9.314432867571732e-06,
      "loss": 0.669,
      "step": 1289
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1647542715072632,
      "learning_rate": 9.31320287463734e-06,
      "loss": 0.5001,
      "step": 1290
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4013057947158813,
      "learning_rate": 9.311971860674014e-06,
      "loss": 0.4773,
      "step": 1291
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0908682346343994,
      "learning_rate": 9.31073982597316e-06,
      "loss": 0.5117,
      "step": 1292
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9733484983444214,
      "learning_rate": 9.309506770826427e-06,
      "loss": 0.4694,
      "step": 1293
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8697170615196228,
      "learning_rate": 9.30827269552571e-06,
      "loss": 0.4982,
      "step": 1294
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.724470853805542,
      "learning_rate": 9.307037600363133e-06,
      "loss": 0.4657,
      "step": 1295
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4599709510803223,
      "learning_rate": 9.305801485631076e-06,
      "loss": 0.5615,
      "step": 1296
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.906561017036438,
      "learning_rate": 9.304564351622156e-06,
      "loss": 0.4247,
      "step": 1297
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4349650144577026,
      "learning_rate": 9.303326198629225e-06,
      "loss": 0.4864,
      "step": 1298
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4270541667938232,
      "learning_rate": 9.302087026945382e-06,
      "loss": 0.4342,
      "step": 1299
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9846651554107666,
      "learning_rate": 9.300846836863968e-06,
      "loss": 0.5267,
      "step": 1300
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0081912279129028,
      "learning_rate": 9.29960562867856e-06,
      "loss": 0.4341,
      "step": 1301
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.017206072807312,
      "learning_rate": 9.29836340268298e-06,
      "loss": 0.5024,
      "step": 1302
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4377521276474,
      "learning_rate": 9.297120159171292e-06,
      "loss": 0.5058,
      "step": 1303
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0645580291748047,
      "learning_rate": 9.295875898437795e-06,
      "loss": 0.4336,
      "step": 1304
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.250229835510254,
      "learning_rate": 9.294630620777035e-06,
      "loss": 0.4417,
      "step": 1305
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3992738723754883,
      "learning_rate": 9.293384326483797e-06,
      "loss": 0.4745,
      "step": 1306
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9582592248916626,
      "learning_rate": 9.292137015853104e-06,
      "loss": 0.5018,
      "step": 1307
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1123452186584473,
      "learning_rate": 9.290888689180223e-06,
      "loss": 0.4468,
      "step": 1308
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3082910776138306,
      "learning_rate": 9.289639346760659e-06,
      "loss": 0.4585,
      "step": 1309
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4158395528793335,
      "learning_rate": 9.288388988890157e-06,
      "loss": 0.5598,
      "step": 1310
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.663513422012329,
      "learning_rate": 9.287137615864709e-06,
      "loss": 0.5089,
      "step": 1311
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4777663946151733,
      "learning_rate": 9.285885227980538e-06,
      "loss": 0.4864,
      "step": 1312
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1187540292739868,
      "learning_rate": 9.284631825534113e-06,
      "loss": 0.4742,
      "step": 1313
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6854199171066284,
      "learning_rate": 9.283377408822141e-06,
      "loss": 0.4986,
      "step": 1314
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9355700612068176,
      "learning_rate": 9.28212197814157e-06,
      "loss": 0.4755,
      "step": 1315
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3756757974624634,
      "learning_rate": 9.280865533789588e-06,
      "loss": 0.5166,
      "step": 1316
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0847245454788208,
      "learning_rate": 9.279608076063622e-06,
      "loss": 0.4285,
      "step": 1317
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8688002824783325,
      "learning_rate": 9.27834960526134e-06,
      "loss": 0.426,
      "step": 1318
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1471587419509888,
      "learning_rate": 9.277090121680649e-06,
      "loss": 0.4572,
      "step": 1319
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.059627652168274,
      "learning_rate": 9.275829625619698e-06,
      "loss": 0.4676,
      "step": 1320
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2502223253250122,
      "learning_rate": 9.274568117376869e-06,
      "loss": 0.4856,
      "step": 1321
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0817372798919678,
      "learning_rate": 9.273305597250797e-06,
      "loss": 0.4927,
      "step": 1322
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.37164306640625,
      "learning_rate": 9.272042065540338e-06,
      "loss": 0.5123,
      "step": 1323
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2675671577453613,
      "learning_rate": 9.270777522544605e-06,
      "loss": 0.5309,
      "step": 1324
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6057231426239014,
      "learning_rate": 9.269511968562939e-06,
      "loss": 0.5248,
      "step": 1325
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.336911678314209,
      "learning_rate": 9.268245403894925e-06,
      "loss": 0.4768,
      "step": 1326
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.408381700515747,
      "learning_rate": 9.266977828840387e-06,
      "loss": 0.5075,
      "step": 1327
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0780293941497803,
      "learning_rate": 9.265709243699388e-06,
      "loss": 0.5217,
      "step": 1328
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8481930494308472,
      "learning_rate": 9.264439648772229e-06,
      "loss": 0.5012,
      "step": 1329
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0078091621398926,
      "learning_rate": 9.263169044359449e-06,
      "loss": 0.4675,
      "step": 1330
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1582167148590088,
      "learning_rate": 9.261897430761831e-06,
      "loss": 0.4414,
      "step": 1331
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0710002183914185,
      "learning_rate": 9.26062480828039e-06,
      "loss": 0.451,
      "step": 1332
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4546737670898438,
      "learning_rate": 9.259351177216386e-06,
      "loss": 0.4602,
      "step": 1333
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0661940574645996,
      "learning_rate": 9.258076537871317e-06,
      "loss": 0.4264,
      "step": 1334
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.984075665473938,
      "learning_rate": 9.256800890546913e-06,
      "loss": 0.5063,
      "step": 1335
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3930745124816895,
      "learning_rate": 9.25552423554515e-06,
      "loss": 0.4885,
      "step": 1336
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7523460388183594,
      "learning_rate": 9.25424657316824e-06,
      "loss": 0.4882,
      "step": 1337
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2597038745880127,
      "learning_rate": 9.252967903718634e-06,
      "loss": 0.7691,
      "step": 1338
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1262694597244263,
      "learning_rate": 9.25168822749902e-06,
      "loss": 0.5184,
      "step": 1339
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9295517206192017,
      "learning_rate": 9.250407544812325e-06,
      "loss": 0.5389,
      "step": 1340
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2426953315734863,
      "learning_rate": 9.249125855961714e-06,
      "loss": 0.5337,
      "step": 1341
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.202747106552124,
      "learning_rate": 9.247843161250593e-06,
      "loss": 0.7553,
      "step": 1342
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1080622673034668,
      "learning_rate": 9.246559460982603e-06,
      "loss": 0.4556,
      "step": 1343
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3791526556015015,
      "learning_rate": 9.245274755461621e-06,
      "loss": 0.441,
      "step": 1344
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1935701370239258,
      "learning_rate": 9.243989044991768e-06,
      "loss": 0.4657,
      "step": 1345
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1225513219833374,
      "learning_rate": 9.242702329877397e-06,
      "loss": 0.5323,
      "step": 1346
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5724573135375977,
      "learning_rate": 9.241414610423104e-06,
      "loss": 0.5343,
      "step": 1347
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.147439956665039,
      "learning_rate": 9.240125886933718e-06,
      "loss": 0.5121,
      "step": 1348
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5322060585021973,
      "learning_rate": 9.23883615971431e-06,
      "loss": 0.4605,
      "step": 1349
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.252120852470398,
      "learning_rate": 9.237545429070185e-06,
      "loss": 0.5243,
      "step": 1350
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2689242362976074,
      "learning_rate": 9.236253695306888e-06,
      "loss": 0.5082,
      "step": 1351
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2545092105865479,
      "learning_rate": 9.234960958730199e-06,
      "loss": 0.5085,
      "step": 1352
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.177768588066101,
      "learning_rate": 9.233667219646139e-06,
      "loss": 0.4039,
      "step": 1353
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0727647542953491,
      "learning_rate": 9.23237247836096e-06,
      "loss": 0.4389,
      "step": 1354
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0129882097244263,
      "learning_rate": 9.231076735181162e-06,
      "loss": 0.5115,
      "step": 1355
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.15574049949646,
      "learning_rate": 9.22977999041347e-06,
      "loss": 0.5183,
      "step": 1356
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4461718797683716,
      "learning_rate": 9.228482244364854e-06,
      "loss": 0.5106,
      "step": 1357
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5371363162994385,
      "learning_rate": 9.227183497342519e-06,
      "loss": 0.506,
      "step": 1358
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.27281653881073,
      "learning_rate": 9.225883749653905e-06,
      "loss": 0.3644,
      "step": 1359
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3568626642227173,
      "learning_rate": 9.224583001606691e-06,
      "loss": 0.5035,
      "step": 1360
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1724951267242432,
      "learning_rate": 9.223281253508796e-06,
      "loss": 0.4579,
      "step": 1361
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.494379997253418,
      "learning_rate": 9.221978505668365e-06,
      "loss": 0.5711,
      "step": 1362
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.425607681274414,
      "learning_rate": 9.220674758393792e-06,
      "loss": 0.7752,
      "step": 1363
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4165074825286865,
      "learning_rate": 9.219370011993703e-06,
      "loss": 0.4684,
      "step": 1364
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.207376480102539,
      "learning_rate": 9.218064266776958e-06,
      "loss": 0.453,
      "step": 1365
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3844774961471558,
      "learning_rate": 9.216757523052652e-06,
      "loss": 0.458,
      "step": 1366
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9314460754394531,
      "learning_rate": 9.215449781130127e-06,
      "loss": 0.5127,
      "step": 1367
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0959292650222778,
      "learning_rate": 9.214141041318947e-06,
      "loss": 0.442,
      "step": 1368
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.044449806213379,
      "learning_rate": 9.212831303928924e-06,
      "loss": 0.4623,
      "step": 1369
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0552512407302856,
      "learning_rate": 9.211520569270098e-06,
      "loss": 0.5056,
      "step": 1370
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1228973865509033,
      "learning_rate": 9.210208837652749e-06,
      "loss": 0.3808,
      "step": 1371
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1065346002578735,
      "learning_rate": 9.208896109387395e-06,
      "loss": 0.5114,
      "step": 1372
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.387096881866455,
      "learning_rate": 9.207582384784786e-06,
      "loss": 0.4973,
      "step": 1373
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1430020332336426,
      "learning_rate": 9.206267664155906e-06,
      "loss": 0.5344,
      "step": 1374
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3504732847213745,
      "learning_rate": 9.204951947811984e-06,
      "loss": 0.4634,
      "step": 1375
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3673136234283447,
      "learning_rate": 9.203635236064473e-06,
      "loss": 0.5124,
      "step": 1376
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0557514429092407,
      "learning_rate": 9.20231752922507e-06,
      "loss": 0.5013,
      "step": 1377
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.993939757347107,
      "learning_rate": 9.200998827605705e-06,
      "loss": 0.5219,
      "step": 1378
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1924980878829956,
      "learning_rate": 9.199679131518542e-06,
      "loss": 0.4657,
      "step": 1379
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4015299081802368,
      "learning_rate": 9.198358441275983e-06,
      "loss": 0.5826,
      "step": 1380
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0694342851638794,
      "learning_rate": 9.197036757190665e-06,
      "loss": 0.4422,
      "step": 1381
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.38303542137146,
      "learning_rate": 9.195714079575458e-06,
      "loss": 0.4386,
      "step": 1382
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3897063732147217,
      "learning_rate": 9.194390408743467e-06,
      "loss": 0.5712,
      "step": 1383
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1532644033432007,
      "learning_rate": 9.193065745008036e-06,
      "loss": 0.487,
      "step": 1384
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1991182565689087,
      "learning_rate": 9.191740088682742e-06,
      "loss": 0.4843,
      "step": 1385
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4335813522338867,
      "learning_rate": 9.190413440081395e-06,
      "loss": 0.4259,
      "step": 1386
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.962899684906006,
      "learning_rate": 9.189085799518043e-06,
      "loss": 0.523,
      "step": 1387
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6656094789505005,
      "learning_rate": 9.187757167306967e-06,
      "loss": 0.4343,
      "step": 1388
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2862956523895264,
      "learning_rate": 9.186427543762681e-06,
      "loss": 0.4654,
      "step": 1389
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1667091846466064,
      "learning_rate": 9.18509692919994e-06,
      "loss": 0.48,
      "step": 1390
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.427554726600647,
      "learning_rate": 9.183765323933727e-06,
      "loss": 0.4604,
      "step": 1391
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0739163160324097,
      "learning_rate": 9.18243272827926e-06,
      "loss": 0.4577,
      "step": 1392
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3137047290802002,
      "learning_rate": 9.181099142551998e-06,
      "loss": 0.5104,
      "step": 1393
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1099461317062378,
      "learning_rate": 9.179764567067624e-06,
      "loss": 0.4578,
      "step": 1394
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6596208810806274,
      "learning_rate": 9.178429002142066e-06,
      "loss": 0.4735,
      "step": 1395
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.061966061592102,
      "learning_rate": 9.17709244809148e-06,
      "loss": 0.5218,
      "step": 1396
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1227784156799316,
      "learning_rate": 9.175754905232258e-06,
      "loss": 0.4283,
      "step": 1397
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2313727140426636,
      "learning_rate": 9.17441637388102e-06,
      "loss": 0.542,
      "step": 1398
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3092902898788452,
      "learning_rate": 9.173076854354634e-06,
      "loss": 0.7691,
      "step": 1399
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1156868934631348,
      "learning_rate": 9.171736346970188e-06,
      "loss": 0.5125,
      "step": 1400
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.46363365650177,
      "learning_rate": 9.17039485204501e-06,
      "loss": 0.5028,
      "step": 1401
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2311121225357056,
      "learning_rate": 9.169052369896663e-06,
      "loss": 0.4892,
      "step": 1402
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3876237869262695,
      "learning_rate": 9.16770890084294e-06,
      "loss": 0.5663,
      "step": 1403
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2331243753433228,
      "learning_rate": 9.166364445201868e-06,
      "loss": 0.5452,
      "step": 1404
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8242319822311401,
      "learning_rate": 9.165019003291712e-06,
      "loss": 0.4382,
      "step": 1405
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4716578722000122,
      "learning_rate": 9.163672575430965e-06,
      "loss": 0.4928,
      "step": 1406
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3819959163665771,
      "learning_rate": 9.162325161938356e-06,
      "loss": 0.4856,
      "step": 1407
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1180404424667358,
      "learning_rate": 9.16097676313285e-06,
      "loss": 0.57,
      "step": 1408
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8289717435836792,
      "learning_rate": 9.15962737933364e-06,
      "loss": 0.4993,
      "step": 1409
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6572545766830444,
      "learning_rate": 9.158277010860152e-06,
      "loss": 0.4892,
      "step": 1410
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.453958988189697,
      "learning_rate": 9.156925658032052e-06,
      "loss": 0.4712,
      "step": 1411
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2661874294281006,
      "learning_rate": 9.155573321169234e-06,
      "loss": 0.5267,
      "step": 1412
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.451675295829773,
      "learning_rate": 9.154220000591821e-06,
      "loss": 0.5227,
      "step": 1413
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2480705976486206,
      "learning_rate": 9.152865696620178e-06,
      "loss": 0.5521,
      "step": 1414
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.046258568763733,
      "learning_rate": 9.151510409574897e-06,
      "loss": 0.388,
      "step": 1415
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3025764226913452,
      "learning_rate": 9.150154139776806e-06,
      "loss": 0.4366,
      "step": 1416
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1798127889633179,
      "learning_rate": 9.148796887546959e-06,
      "loss": 0.7001,
      "step": 1417
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4341027736663818,
      "learning_rate": 9.14743865320665e-06,
      "loss": 0.5079,
      "step": 1418
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2189345359802246,
      "learning_rate": 9.146079437077401e-06,
      "loss": 0.4426,
      "step": 1419
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.226291298866272,
      "learning_rate": 9.144719239480972e-06,
      "loss": 0.432,
      "step": 1420
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2951298952102661,
      "learning_rate": 9.143358060739347e-06,
      "loss": 0.4241,
      "step": 1421
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.025933861732483,
      "learning_rate": 9.141995901174749e-06,
      "loss": 0.4512,
      "step": 1422
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5279115438461304,
      "learning_rate": 9.14063276110963e-06,
      "loss": 0.5469,
      "step": 1423
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0372607707977295,
      "learning_rate": 9.139268640866673e-06,
      "loss": 0.5173,
      "step": 1424
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.5869662761688232,
      "learning_rate": 9.137903540768798e-06,
      "loss": 0.4264,
      "step": 1425
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6076130867004395,
      "learning_rate": 9.13653746113915e-06,
      "loss": 0.6071,
      "step": 1426
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3337634801864624,
      "learning_rate": 9.135170402301117e-06,
      "loss": 0.4334,
      "step": 1427
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3680330514907837,
      "learning_rate": 9.133802364578304e-06,
      "loss": 0.3856,
      "step": 1428
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1019099950790405,
      "learning_rate": 9.132433348294559e-06,
      "loss": 0.4692,
      "step": 1429
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1425600051879883,
      "learning_rate": 9.131063353773956e-06,
      "loss": 0.4581,
      "step": 1430
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.48952317237854,
      "learning_rate": 9.129692381340807e-06,
      "loss": 0.5121,
      "step": 1431
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3911287784576416,
      "learning_rate": 9.128320431319644e-06,
      "loss": 0.7607,
      "step": 1432
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2016664743423462,
      "learning_rate": 9.126947504035244e-06,
      "loss": 0.5263,
      "step": 1433
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0781075954437256,
      "learning_rate": 9.125573599812604e-06,
      "loss": 0.457,
      "step": 1434
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.341780662536621,
      "learning_rate": 9.124198718976959e-06,
      "loss": 0.4784,
      "step": 1435
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4602785110473633,
      "learning_rate": 9.122822861853774e-06,
      "loss": 0.4898,
      "step": 1436
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.607614517211914,
      "learning_rate": 9.121446028768744e-06,
      "loss": 0.4556,
      "step": 1437
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2711372375488281,
      "learning_rate": 9.120068220047793e-06,
      "loss": 0.5436,
      "step": 1438
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4023584127426147,
      "learning_rate": 9.11868943601708e-06,
      "loss": 0.5458,
      "step": 1439
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3307284116744995,
      "learning_rate": 9.117309677002994e-06,
      "loss": 0.4439,
      "step": 1440
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.551137089729309,
      "learning_rate": 9.115928943332152e-06,
      "loss": 0.5512,
      "step": 1441
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.284578800201416,
      "learning_rate": 9.114547235331405e-06,
      "loss": 0.528,
      "step": 1442
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5121184587478638,
      "learning_rate": 9.113164553327835e-06,
      "loss": 0.4758,
      "step": 1443
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5879017114639282,
      "learning_rate": 9.11178089764875e-06,
      "loss": 0.5248,
      "step": 1444
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4768056869506836,
      "learning_rate": 9.110396268621692e-06,
      "loss": 0.4784,
      "step": 1445
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0894583463668823,
      "learning_rate": 9.109010666574434e-06,
      "loss": 0.5004,
      "step": 1446
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1777714490890503,
      "learning_rate": 9.107624091834976e-06,
      "loss": 0.515,
      "step": 1447
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.136574625968933,
      "learning_rate": 9.106236544731553e-06,
      "loss": 0.3866,
      "step": 1448
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2013198137283325,
      "learning_rate": 9.104848025592627e-06,
      "loss": 0.5532,
      "step": 1449
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2960689067840576,
      "learning_rate": 9.103458534746889e-06,
      "loss": 0.535,
      "step": 1450
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2680829763412476,
      "learning_rate": 9.102068072523264e-06,
      "loss": 0.4508,
      "step": 1451
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0560429096221924,
      "learning_rate": 9.100676639250903e-06,
      "loss": 0.4784,
      "step": 1452
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6553733348846436,
      "learning_rate": 9.099284235259189e-06,
      "loss": 0.5478,
      "step": 1453
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0860612392425537,
      "learning_rate": 9.097890860877732e-06,
      "loss": 0.528,
      "step": 1454
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1591565608978271,
      "learning_rate": 9.09649651643638e-06,
      "loss": 0.49,
      "step": 1455
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.9171533584594727,
      "learning_rate": 9.095101202265198e-06,
      "loss": 0.4981,
      "step": 1456
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.149023413658142,
      "learning_rate": 9.093704918694491e-06,
      "loss": 0.4913,
      "step": 1457
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2844830751419067,
      "learning_rate": 9.09230766605479e-06,
      "loss": 0.4847,
      "step": 1458
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3575628995895386,
      "learning_rate": 9.090909444676852e-06,
      "loss": 0.5047,
      "step": 1459
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2239633798599243,
      "learning_rate": 9.08951025489167e-06,
      "loss": 0.5272,
      "step": 1460
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0344713926315308,
      "learning_rate": 9.088110097030461e-06,
      "loss": 0.5109,
      "step": 1461
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.134347438812256,
      "learning_rate": 9.086708971424671e-06,
      "loss": 0.4989,
      "step": 1462
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2113546133041382,
      "learning_rate": 9.085306878405981e-06,
      "loss": 0.4074,
      "step": 1463
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2509887218475342,
      "learning_rate": 9.083903818306294e-06,
      "loss": 0.4575,
      "step": 1464
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.160709023475647,
      "learning_rate": 9.082499791457744e-06,
      "loss": 0.455,
      "step": 1465
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2435643672943115,
      "learning_rate": 9.081094798192698e-06,
      "loss": 0.4546,
      "step": 1466
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4810653924942017,
      "learning_rate": 9.079688838843746e-06,
      "loss": 0.4722,
      "step": 1467
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5983147621154785,
      "learning_rate": 9.07828191374371e-06,
      "loss": 0.5285,
      "step": 1468
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4401435852050781,
      "learning_rate": 9.076874023225638e-06,
      "loss": 0.4907,
      "step": 1469
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3354463577270508,
      "learning_rate": 9.075465167622812e-06,
      "loss": 0.432,
      "step": 1470
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.076974868774414,
      "learning_rate": 9.074055347268735e-06,
      "loss": 0.4722,
      "step": 1471
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3017321825027466,
      "learning_rate": 9.072644562497145e-06,
      "loss": 0.7463,
      "step": 1472
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2094347476959229,
      "learning_rate": 9.071232813642005e-06,
      "loss": 0.4682,
      "step": 1473
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2281054258346558,
      "learning_rate": 9.069820101037506e-06,
      "loss": 0.4358,
      "step": 1474
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6417664289474487,
      "learning_rate": 9.068406425018066e-06,
      "loss": 0.4616,
      "step": 1475
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.372689127922058,
      "learning_rate": 9.066991785918334e-06,
      "loss": 0.7981,
      "step": 1476
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.267285943031311,
      "learning_rate": 9.065576184073188e-06,
      "loss": 0.445,
      "step": 1477
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3550746440887451,
      "learning_rate": 9.06415961981773e-06,
      "loss": 0.4293,
      "step": 1478
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.402666449546814,
      "learning_rate": 9.062742093487291e-06,
      "loss": 0.5377,
      "step": 1479
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6844491958618164,
      "learning_rate": 9.061323605417431e-06,
      "loss": 0.5161,
      "step": 1480
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.043405055999756,
      "learning_rate": 9.059904155943938e-06,
      "loss": 0.5434,
      "step": 1481
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4057122468948364,
      "learning_rate": 9.058483745402826e-06,
      "loss": 0.4934,
      "step": 1482
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.408704161643982,
      "learning_rate": 9.057062374130334e-06,
      "loss": 0.4477,
      "step": 1483
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5187108516693115,
      "learning_rate": 9.055640042462937e-06,
      "loss": 0.5138,
      "step": 1484
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4802265167236328,
      "learning_rate": 9.054216750737328e-06,
      "loss": 0.4829,
      "step": 1485
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2806296348571777,
      "learning_rate": 9.052792499290432e-06,
      "loss": 0.4944,
      "step": 1486
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2645105123519897,
      "learning_rate": 9.0513672884594e-06,
      "loss": 0.5113,
      "step": 1487
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1205928325653076,
      "learning_rate": 9.04994111858161e-06,
      "loss": 0.5144,
      "step": 1488
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1377983093261719,
      "learning_rate": 9.04851398999467e-06,
      "loss": 0.5201,
      "step": 1489
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9151362180709839,
      "learning_rate": 9.04708590303641e-06,
      "loss": 0.478,
      "step": 1490
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.296416997909546,
      "learning_rate": 9.04565685804489e-06,
      "loss": 0.4448,
      "step": 1491
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.326406717300415,
      "learning_rate": 9.044226855358395e-06,
      "loss": 0.4832,
      "step": 1492
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1984174251556396,
      "learning_rate": 9.04279589531544e-06,
      "loss": 0.4883,
      "step": 1493
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5652741193771362,
      "learning_rate": 9.041363978254763e-06,
      "loss": 0.4818,
      "step": 1494
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.207844853401184,
      "learning_rate": 9.039931104515329e-06,
      "loss": 0.4892,
      "step": 1495
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6331820487976074,
      "learning_rate": 9.038497274436332e-06,
      "loss": 0.468,
      "step": 1496
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.799936056137085,
      "learning_rate": 9.03706248835719e-06,
      "loss": 0.4676,
      "step": 1497
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8609766960144043,
      "learning_rate": 9.035626746617547e-06,
      "loss": 0.5514,
      "step": 1498
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5977699756622314,
      "learning_rate": 9.034190049557277e-06,
      "loss": 0.544,
      "step": 1499
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1579760313034058,
      "learning_rate": 9.032752397516475e-06,
      "loss": 0.7223,
      "step": 1500
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1865406036376953,
      "learning_rate": 9.031313790835466e-06,
      "loss": 0.7553,
      "step": 1501
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9807384014129639,
      "learning_rate": 9.029874229854796e-06,
      "loss": 0.4287,
      "step": 1502
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4645668268203735,
      "learning_rate": 9.028433714915247e-06,
      "loss": 0.5337,
      "step": 1503
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9681477546691895,
      "learning_rate": 9.026992246357815e-06,
      "loss": 0.5072,
      "step": 1504
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2410995960235596,
      "learning_rate": 9.025549824523729e-06,
      "loss": 0.481,
      "step": 1505
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.292168140411377,
      "learning_rate": 9.02410644975444e-06,
      "loss": 0.4279,
      "step": 1506
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6217055320739746,
      "learning_rate": 9.022662122391628e-06,
      "loss": 0.5426,
      "step": 1507
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3946999311447144,
      "learning_rate": 9.021216842777196e-06,
      "loss": 0.4971,
      "step": 1508
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2912914752960205,
      "learning_rate": 9.019770611253272e-06,
      "loss": 0.4962,
      "step": 1509
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4411730766296387,
      "learning_rate": 9.018323428162211e-06,
      "loss": 0.4776,
      "step": 1510
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4840641021728516,
      "learning_rate": 9.016875293846597e-06,
      "loss": 0.5824,
      "step": 1511
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.416359782218933,
      "learning_rate": 9.015426208649227e-06,
      "loss": 0.614,
      "step": 1512
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4225784540176392,
      "learning_rate": 9.013976172913136e-06,
      "loss": 0.4246,
      "step": 1513
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0273070335388184,
      "learning_rate": 9.01252518698158e-06,
      "loss": 0.5294,
      "step": 1514
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2110759019851685,
      "learning_rate": 9.011073251198038e-06,
      "loss": 0.5134,
      "step": 1515
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2944859266281128,
      "learning_rate": 9.009620365906211e-06,
      "loss": 0.4429,
      "step": 1516
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1952056884765625,
      "learning_rate": 9.008166531450031e-06,
      "loss": 0.4466,
      "step": 1517
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4049699306488037,
      "learning_rate": 9.006711748173655e-06,
      "loss": 0.5779,
      "step": 1518
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.394493818283081,
      "learning_rate": 9.005256016421458e-06,
      "loss": 0.4401,
      "step": 1519
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1224948167800903,
      "learning_rate": 9.003799336538046e-06,
      "loss": 0.4181,
      "step": 1520
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7079551219940186,
      "learning_rate": 9.002341708868245e-06,
      "loss": 0.4774,
      "step": 1521
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1005505323410034,
      "learning_rate": 9.000883133757108e-06,
      "loss": 0.4615,
      "step": 1522
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3261488676071167,
      "learning_rate": 8.99942361154991e-06,
      "loss": 0.4485,
      "step": 1523
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.704415202140808,
      "learning_rate": 8.997963142592154e-06,
      "loss": 0.5192,
      "step": 1524
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0590403079986572,
      "learning_rate": 8.996501727229565e-06,
      "loss": 0.4354,
      "step": 1525
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.651944875717163,
      "learning_rate": 8.99503936580809e-06,
      "loss": 0.5212,
      "step": 1526
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2590638399124146,
      "learning_rate": 8.9935760586739e-06,
      "loss": 0.521,
      "step": 1527
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2009117603302002,
      "learning_rate": 8.992111806173396e-06,
      "loss": 0.48,
      "step": 1528
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7046018838882446,
      "learning_rate": 8.990646608653196e-06,
      "loss": 0.5673,
      "step": 1529
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.5838887691497803,
      "learning_rate": 8.989180466460143e-06,
      "loss": 0.4501,
      "step": 1530
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.702834963798523,
      "learning_rate": 8.987713379941307e-06,
      "loss": 0.514,
      "step": 1531
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4543217420578003,
      "learning_rate": 8.98624534944398e-06,
      "loss": 0.4693,
      "step": 1532
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3346917629241943,
      "learning_rate": 8.984776375315674e-06,
      "loss": 0.4008,
      "step": 1533
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2121063470840454,
      "learning_rate": 8.983306457904128e-06,
      "loss": 0.5489,
      "step": 1534
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0540435314178467,
      "learning_rate": 8.981835597557304e-06,
      "loss": 0.4353,
      "step": 1535
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1566270589828491,
      "learning_rate": 8.980363794623385e-06,
      "loss": 0.5414,
      "step": 1536
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1454284191131592,
      "learning_rate": 8.978891049450783e-06,
      "loss": 0.5105,
      "step": 1537
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2002859115600586,
      "learning_rate": 8.977417362388125e-06,
      "loss": 0.4986,
      "step": 1538
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2543859481811523,
      "learning_rate": 8.975942733784266e-06,
      "loss": 0.4658,
      "step": 1539
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7030034065246582,
      "learning_rate": 8.97446716398828e-06,
      "loss": 0.5881,
      "step": 1540
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.199131965637207,
      "learning_rate": 8.972990653349472e-06,
      "loss": 0.4422,
      "step": 1541
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2616376876831055,
      "learning_rate": 8.971513202217359e-06,
      "loss": 0.4563,
      "step": 1542
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2874641418457031,
      "learning_rate": 8.970034810941688e-06,
      "loss": 0.434,
      "step": 1543
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7994799613952637,
      "learning_rate": 8.968555479872428e-06,
      "loss": 0.3758,
      "step": 1544
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3090671300888062,
      "learning_rate": 8.967075209359766e-06,
      "loss": 0.4256,
      "step": 1545
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1680783033370972,
      "learning_rate": 8.965593999754116e-06,
      "loss": 0.434,
      "step": 1546
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1954636573791504,
      "learning_rate": 8.96411185140611e-06,
      "loss": 0.49,
      "step": 1547
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2646323442459106,
      "learning_rate": 8.962628764666609e-06,
      "loss": 0.461,
      "step": 1548
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1986746788024902,
      "learning_rate": 8.961144739886688e-06,
      "loss": 0.5042,
      "step": 1549
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3155863285064697,
      "learning_rate": 8.95965977741765e-06,
      "loss": 0.4722,
      "step": 1550
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2459204196929932,
      "learning_rate": 8.958173877611017e-06,
      "loss": 0.5348,
      "step": 1551
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.549199104309082,
      "learning_rate": 8.956687040818534e-06,
      "loss": 0.5183,
      "step": 1552
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2874850034713745,
      "learning_rate": 8.955199267392168e-06,
      "loss": 0.5093,
      "step": 1553
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.499000072479248,
      "learning_rate": 8.953710557684108e-06,
      "loss": 0.4238,
      "step": 1554
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9923268556594849,
      "learning_rate": 8.952220912046763e-06,
      "loss": 0.4406,
      "step": 1555
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2316783666610718,
      "learning_rate": 8.950730330832762e-06,
      "loss": 0.5227,
      "step": 1556
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2251300811767578,
      "learning_rate": 8.949238814394963e-06,
      "loss": 0.4466,
      "step": 1557
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4683188199996948,
      "learning_rate": 8.947746363086438e-06,
      "loss": 0.5056,
      "step": 1558
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3031052350997925,
      "learning_rate": 8.946252977260481e-06,
      "loss": 0.4932,
      "step": 1559
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2308781147003174,
      "learning_rate": 8.944758657270614e-06,
      "loss": 0.4673,
      "step": 1560
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.20644211769104,
      "learning_rate": 8.943263403470568e-06,
      "loss": 0.7556,
      "step": 1561
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7015618085861206,
      "learning_rate": 8.941767216214307e-06,
      "loss": 0.5049,
      "step": 1562
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.35291588306427,
      "learning_rate": 8.94027009585601e-06,
      "loss": 0.484,
      "step": 1563
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2860400676727295,
      "learning_rate": 8.938772042750078e-06,
      "loss": 0.4918,
      "step": 1564
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.328425645828247,
      "learning_rate": 8.937273057251133e-06,
      "loss": 0.4782,
      "step": 1565
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2006133794784546,
      "learning_rate": 8.935773139714015e-06,
      "loss": 0.5156,
      "step": 1566
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.421360969543457,
      "learning_rate": 8.93427229049379e-06,
      "loss": 0.4804,
      "step": 1567
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3501288890838623,
      "learning_rate": 8.93277050994574e-06,
      "loss": 0.6031,
      "step": 1568
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3540995121002197,
      "learning_rate": 8.931267798425372e-06,
      "loss": 0.5125,
      "step": 1569
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2076793909072876,
      "learning_rate": 8.929764156288406e-06,
      "loss": 0.4918,
      "step": 1570
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.239152431488037,
      "learning_rate": 8.92825958389079e-06,
      "loss": 0.4896,
      "step": 1571
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2774971723556519,
      "learning_rate": 8.926754081588688e-06,
      "loss": 0.5456,
      "step": 1572
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4798909425735474,
      "learning_rate": 8.925247649738487e-06,
      "loss": 0.5256,
      "step": 1573
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.324388861656189,
      "learning_rate": 8.923740288696789e-06,
      "loss": 0.5045,
      "step": 1574
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9793827533721924,
      "learning_rate": 8.92223199882042e-06,
      "loss": 0.461,
      "step": 1575
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1740437746047974,
      "learning_rate": 8.920722780466426e-06,
      "loss": 0.4582,
      "step": 1576
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2592874765396118,
      "learning_rate": 8.919212633992071e-06,
      "loss": 0.5495,
      "step": 1577
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7685836553573608,
      "learning_rate": 8.917701559754842e-06,
      "loss": 0.4637,
      "step": 1578
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2285405397415161,
      "learning_rate": 8.916189558112441e-06,
      "loss": 0.5213,
      "step": 1579
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3882883787155151,
      "learning_rate": 8.914676629422789e-06,
      "loss": 0.4726,
      "step": 1580
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.157157063484192,
      "learning_rate": 8.913162774044034e-06,
      "loss": 0.4386,
      "step": 1581
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2635172605514526,
      "learning_rate": 8.911647992334536e-06,
      "loss": 0.4901,
      "step": 1582
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3333255052566528,
      "learning_rate": 8.910132284652876e-06,
      "loss": 0.4721,
      "step": 1583
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4718401432037354,
      "learning_rate": 8.908615651357858e-06,
      "loss": 0.5185,
      "step": 1584
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1600844860076904,
      "learning_rate": 8.9070980928085e-06,
      "loss": 0.4679,
      "step": 1585
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1148655414581299,
      "learning_rate": 8.905579609364041e-06,
      "loss": 0.4639,
      "step": 1586
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5086554288864136,
      "learning_rate": 8.904060201383939e-06,
      "loss": 0.5407,
      "step": 1587
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2343019247055054,
      "learning_rate": 8.90253986922787e-06,
      "loss": 0.4971,
      "step": 1588
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9727808833122253,
      "learning_rate": 8.901018613255734e-06,
      "loss": 0.4626,
      "step": 1589
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3566926717758179,
      "learning_rate": 8.899496433827642e-06,
      "loss": 0.4332,
      "step": 1590
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3003604412078857,
      "learning_rate": 8.897973331303927e-06,
      "loss": 0.4735,
      "step": 1591
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2784947156906128,
      "learning_rate": 8.89644930604514e-06,
      "loss": 0.4957,
      "step": 1592
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3241283893585205,
      "learning_rate": 8.894924358412052e-06,
      "loss": 0.5224,
      "step": 1593
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4756275415420532,
      "learning_rate": 8.89339848876565e-06,
      "loss": 0.4716,
      "step": 1594
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2889089584350586,
      "learning_rate": 8.891871697467143e-06,
      "loss": 0.5083,
      "step": 1595
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9247274994850159,
      "learning_rate": 8.890343984877954e-06,
      "loss": 0.4653,
      "step": 1596
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3349230289459229,
      "learning_rate": 8.888815351359727e-06,
      "loss": 0.4006,
      "step": 1597
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.607163667678833,
      "learning_rate": 8.887285797274321e-06,
      "loss": 0.4707,
      "step": 1598
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5063923597335815,
      "learning_rate": 8.885755322983816e-06,
      "loss": 0.4919,
      "step": 1599
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5224424600601196,
      "learning_rate": 8.884223928850509e-06,
      "loss": 0.5774,
      "step": 1600
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3497536182403564,
      "learning_rate": 8.882691615236912e-06,
      "loss": 0.4485,
      "step": 1601
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.0983352661132812,
      "learning_rate": 8.88115838250576e-06,
      "loss": 0.4809,
      "step": 1602
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2391722202301025,
      "learning_rate": 8.87962423102e-06,
      "loss": 0.5121,
      "step": 1603
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.62857186794281,
      "learning_rate": 8.8780891611428e-06,
      "loss": 0.4363,
      "step": 1604
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2763535976409912,
      "learning_rate": 8.876553173237544e-06,
      "loss": 0.4736,
      "step": 1605
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4654731750488281,
      "learning_rate": 8.875016267667833e-06,
      "loss": 0.4703,
      "step": 1606
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3117969036102295,
      "learning_rate": 8.873478444797489e-06,
      "loss": 0.4773,
      "step": 1607
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2231557369232178,
      "learning_rate": 8.871939704990547e-06,
      "loss": 0.7484,
      "step": 1608
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.394651174545288,
      "learning_rate": 8.870400048611259e-06,
      "loss": 0.5032,
      "step": 1609
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2954856157302856,
      "learning_rate": 8.868859476024095e-06,
      "loss": 0.5103,
      "step": 1610
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0747073888778687,
      "learning_rate": 8.867317987593744e-06,
      "loss": 0.4811,
      "step": 1611
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3959763050079346,
      "learning_rate": 8.865775583685107e-06,
      "loss": 0.4691,
      "step": 1612
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5230220556259155,
      "learning_rate": 8.864232264663306e-06,
      "loss": 0.4892,
      "step": 1613
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2183256149291992,
      "learning_rate": 8.86268803089368e-06,
      "loss": 0.4613,
      "step": 1614
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.298967957496643,
      "learning_rate": 8.86114288274178e-06,
      "loss": 0.4735,
      "step": 1615
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9902824759483337,
      "learning_rate": 8.859596820573378e-06,
      "loss": 0.5017,
      "step": 1616
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3703025579452515,
      "learning_rate": 8.858049844754457e-06,
      "loss": 0.524,
      "step": 1617
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0587749481201172,
      "learning_rate": 8.856501955651225e-06,
      "loss": 0.6894,
      "step": 1618
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4492316246032715,
      "learning_rate": 8.854953153630097e-06,
      "loss": 0.5053,
      "step": 1619
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0732330083847046,
      "learning_rate": 8.853403439057711e-06,
      "loss": 0.4584,
      "step": 1620
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3474467992782593,
      "learning_rate": 8.851852812300915e-06,
      "loss": 0.4766,
      "step": 1621
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2875956296920776,
      "learning_rate": 8.850301273726781e-06,
      "loss": 0.4453,
      "step": 1622
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5031967163085938,
      "learning_rate": 8.848748823702587e-06,
      "loss": 0.5236,
      "step": 1623
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3352466821670532,
      "learning_rate": 8.847195462595833e-06,
      "loss": 0.396,
      "step": 1624
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8940384984016418,
      "learning_rate": 8.845641190774232e-06,
      "loss": 0.4939,
      "step": 1625
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3878642320632935,
      "learning_rate": 8.844086008605718e-06,
      "loss": 0.4992,
      "step": 1626
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2339881658554077,
      "learning_rate": 8.842529916458436e-06,
      "loss": 0.5366,
      "step": 1627
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7654309272766113,
      "learning_rate": 8.840972914700743e-06,
      "loss": 0.5156,
      "step": 1628
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4447791576385498,
      "learning_rate": 8.839415003701217e-06,
      "loss": 0.4433,
      "step": 1629
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1067687273025513,
      "learning_rate": 8.83785618382865e-06,
      "loss": 0.4742,
      "step": 1630
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.161118268966675,
      "learning_rate": 8.83629645545205e-06,
      "loss": 0.5383,
      "step": 1631
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3631919622421265,
      "learning_rate": 8.834735818940635e-06,
      "loss": 0.5368,
      "step": 1632
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4417574405670166,
      "learning_rate": 8.833174274663844e-06,
      "loss": 0.4721,
      "step": 1633
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.163559913635254,
      "learning_rate": 8.831611822991329e-06,
      "loss": 0.5311,
      "step": 1634
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2583848237991333,
      "learning_rate": 8.830048464292955e-06,
      "loss": 0.4958,
      "step": 1635
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.3908393383026123,
      "learning_rate": 8.828484198938801e-06,
      "loss": 0.4267,
      "step": 1636
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2242769002914429,
      "learning_rate": 8.826919027299168e-06,
      "loss": 0.4023,
      "step": 1637
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.171545386314392,
      "learning_rate": 8.825352949744561e-06,
      "loss": 0.4839,
      "step": 1638
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5079824924468994,
      "learning_rate": 8.823785966645708e-06,
      "loss": 0.4514,
      "step": 1639
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2612779140472412,
      "learning_rate": 8.822218078373545e-06,
      "loss": 0.7458,
      "step": 1640
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.136016607284546,
      "learning_rate": 8.82064928529923e-06,
      "loss": 0.4274,
      "step": 1641
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2801463603973389,
      "learning_rate": 8.819079587794125e-06,
      "loss": 0.7964,
      "step": 1642
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2872861623764038,
      "learning_rate": 8.817508986229815e-06,
      "loss": 0.5144,
      "step": 1643
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2564078569412231,
      "learning_rate": 8.815937480978094e-06,
      "loss": 0.5014,
      "step": 1644
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6520577669143677,
      "learning_rate": 8.814365072410972e-06,
      "loss": 0.5573,
      "step": 1645
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.527491807937622,
      "learning_rate": 8.812791760900672e-06,
      "loss": 0.4341,
      "step": 1646
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0658671855926514,
      "learning_rate": 8.811217546819632e-06,
      "loss": 0.5366,
      "step": 1647
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3277714252471924,
      "learning_rate": 8.809642430540502e-06,
      "loss": 0.4322,
      "step": 1648
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.494913101196289,
      "learning_rate": 8.808066412436148e-06,
      "loss": 0.4892,
      "step": 1649
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2542338371276855,
      "learning_rate": 8.806489492879644e-06,
      "loss": 0.4544,
      "step": 1650
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1835951805114746,
      "learning_rate": 8.804911672244283e-06,
      "loss": 0.5141,
      "step": 1651
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0843158960342407,
      "learning_rate": 8.803332950903571e-06,
      "loss": 0.4818,
      "step": 1652
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1612098217010498,
      "learning_rate": 8.801753329231223e-06,
      "loss": 0.7495,
      "step": 1653
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0624735355377197,
      "learning_rate": 8.800172807601173e-06,
      "loss": 0.5125,
      "step": 1654
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1802492141723633,
      "learning_rate": 8.79859138638756e-06,
      "loss": 0.5025,
      "step": 1655
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0988136529922485,
      "learning_rate": 8.797009065964748e-06,
      "loss": 0.4587,
      "step": 1656
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2088416814804077,
      "learning_rate": 8.7954258467073e-06,
      "loss": 0.4707,
      "step": 1657
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1802854537963867,
      "learning_rate": 8.793841728990001e-06,
      "loss": 0.4601,
      "step": 1658
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.299805998802185,
      "learning_rate": 8.792256713187847e-06,
      "loss": 0.7202,
      "step": 1659
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1379607915878296,
      "learning_rate": 8.790670799676045e-06,
      "loss": 0.4928,
      "step": 1660
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5471712350845337,
      "learning_rate": 8.789083988830014e-06,
      "loss": 0.4521,
      "step": 1661
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1620336771011353,
      "learning_rate": 8.787496281025387e-06,
      "loss": 0.4534,
      "step": 1662
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3488662242889404,
      "learning_rate": 8.78590767663801e-06,
      "loss": 0.5052,
      "step": 1663
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.276495099067688,
      "learning_rate": 8.784318176043942e-06,
      "loss": 0.5194,
      "step": 1664
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.186112403869629,
      "learning_rate": 8.782727779619449e-06,
      "loss": 0.5266,
      "step": 1665
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0873981714248657,
      "learning_rate": 8.781136487741014e-06,
      "loss": 0.4509,
      "step": 1666
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.22007155418396,
      "learning_rate": 8.779544300785328e-06,
      "loss": 0.4437,
      "step": 1667
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2922933101654053,
      "learning_rate": 8.777951219129301e-06,
      "loss": 0.5214,
      "step": 1668
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2660235166549683,
      "learning_rate": 8.776357243150044e-06,
      "loss": 0.4348,
      "step": 1669
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5630372762680054,
      "learning_rate": 8.774762373224892e-06,
      "loss": 0.4656,
      "step": 1670
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.394683599472046,
      "learning_rate": 8.773166609731383e-06,
      "loss": 0.4209,
      "step": 1671
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1031166315078735,
      "learning_rate": 8.771569953047267e-06,
      "loss": 0.4512,
      "step": 1672
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0591697692871094,
      "learning_rate": 8.769972403550509e-06,
      "loss": 0.4526,
      "step": 1673
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1773309707641602,
      "learning_rate": 8.768373961619283e-06,
      "loss": 0.4692,
      "step": 1674
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0914241075515747,
      "learning_rate": 8.766774627631977e-06,
      "loss": 0.4954,
      "step": 1675
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4480860233306885,
      "learning_rate": 8.765174401967187e-06,
      "loss": 0.5057,
      "step": 1676
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1122498512268066,
      "learning_rate": 8.76357328500372e-06,
      "loss": 0.4374,
      "step": 1677
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1998050212860107,
      "learning_rate": 8.761971277120595e-06,
      "loss": 0.5134,
      "step": 1678
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0212733745574951,
      "learning_rate": 8.760368378697046e-06,
      "loss": 0.4605,
      "step": 1679
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.359039306640625,
      "learning_rate": 8.75876459011251e-06,
      "loss": 0.5414,
      "step": 1680
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0604219436645508,
      "learning_rate": 8.757159911746638e-06,
      "loss": 0.7269,
      "step": 1681
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1070685386657715,
      "learning_rate": 8.755554343979296e-06,
      "loss": 0.5456,
      "step": 1682
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2901084423065186,
      "learning_rate": 8.753947887190555e-06,
      "loss": 0.4396,
      "step": 1683
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4046928882598877,
      "learning_rate": 8.752340541760698e-06,
      "loss": 0.5024,
      "step": 1684
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2829382419586182,
      "learning_rate": 8.750732308070218e-06,
      "loss": 0.4217,
      "step": 1685
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3029420375823975,
      "learning_rate": 8.74912318649982e-06,
      "loss": 0.4488,
      "step": 1686
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9884474277496338,
      "learning_rate": 8.747513177430418e-06,
      "loss": 0.4664,
      "step": 1687
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3274770975112915,
      "learning_rate": 8.745902281243135e-06,
      "loss": 0.482,
      "step": 1688
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0460073947906494,
      "learning_rate": 8.744290498319306e-06,
      "loss": 0.4472,
      "step": 1689
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2251639366149902,
      "learning_rate": 8.742677829040477e-06,
      "loss": 0.4054,
      "step": 1690
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.0276668071746826,
      "learning_rate": 8.741064273788399e-06,
      "loss": 0.4579,
      "step": 1691
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1653025150299072,
      "learning_rate": 8.739449832945036e-06,
      "loss": 0.5107,
      "step": 1692
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2887169122695923,
      "learning_rate": 8.737834506892564e-06,
      "loss": 0.5069,
      "step": 1693
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4243357181549072,
      "learning_rate": 8.736218296013362e-06,
      "loss": 0.4443,
      "step": 1694
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.128261685371399,
      "learning_rate": 8.734601200690024e-06,
      "loss": 0.4721,
      "step": 1695
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.499081015586853,
      "learning_rate": 8.732983221305351e-06,
      "loss": 0.4688,
      "step": 1696
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2730532884597778,
      "learning_rate": 8.731364358242355e-06,
      "loss": 0.452,
      "step": 1697
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0973187685012817,
      "learning_rate": 8.729744611884257e-06,
      "loss": 0.4772,
      "step": 1698
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3321051597595215,
      "learning_rate": 8.728123982614485e-06,
      "loss": 0.469,
      "step": 1699
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5591906309127808,
      "learning_rate": 8.726502470816676e-06,
      "loss": 0.4663,
      "step": 1700
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6738096475601196,
      "learning_rate": 8.72488007687468e-06,
      "loss": 0.5042,
      "step": 1701
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0657684803009033,
      "learning_rate": 8.723256801172549e-06,
      "loss": 0.4638,
      "step": 1702
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8249914646148682,
      "learning_rate": 8.721632644094548e-06,
      "loss": 0.5025,
      "step": 1703
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.257351040840149,
      "learning_rate": 8.720007606025155e-06,
      "loss": 0.4666,
      "step": 1704
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3046300411224365,
      "learning_rate": 8.718381687349048e-06,
      "loss": 0.5034,
      "step": 1705
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5461243391036987,
      "learning_rate": 8.71675488845112e-06,
      "loss": 0.5011,
      "step": 1706
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.525022268295288,
      "learning_rate": 8.715127209716467e-06,
      "loss": 0.4906,
      "step": 1707
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2450001239776611,
      "learning_rate": 8.713498651530394e-06,
      "loss": 0.4195,
      "step": 1708
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8616265058517456,
      "learning_rate": 8.711869214278423e-06,
      "loss": 0.5007,
      "step": 1709
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.017082691192627,
      "learning_rate": 8.71023889834627e-06,
      "loss": 0.3982,
      "step": 1710
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2489980459213257,
      "learning_rate": 8.708607704119872e-06,
      "loss": 0.5305,
      "step": 1711
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3640764951705933,
      "learning_rate": 8.706975631985364e-06,
      "loss": 0.4979,
      "step": 1712
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1408562660217285,
      "learning_rate": 8.705342682329096e-06,
      "loss": 0.4685,
      "step": 1713
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3487423658370972,
      "learning_rate": 8.703708855537622e-06,
      "loss": 0.4962,
      "step": 1714
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2408159971237183,
      "learning_rate": 8.7020741519977e-06,
      "loss": 0.5185,
      "step": 1715
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5815576314926147,
      "learning_rate": 8.700438572096308e-06,
      "loss": 0.4588,
      "step": 1716
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3383899927139282,
      "learning_rate": 8.698802116220617e-06,
      "loss": 0.5045,
      "step": 1717
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.191696286201477,
      "learning_rate": 8.697164784758015e-06,
      "loss": 0.4652,
      "step": 1718
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3677645921707153,
      "learning_rate": 8.69552657809609e-06,
      "loss": 0.5084,
      "step": 1719
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2709896564483643,
      "learning_rate": 8.693887496622645e-06,
      "loss": 0.4955,
      "step": 1720
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.344092845916748,
      "learning_rate": 8.692247540725684e-06,
      "loss": 0.4739,
      "step": 1721
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.410568356513977,
      "learning_rate": 8.690606710793423e-06,
      "loss": 0.6327,
      "step": 1722
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6854530572891235,
      "learning_rate": 8.688965007214278e-06,
      "loss": 0.4229,
      "step": 1723
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2597332000732422,
      "learning_rate": 8.68732243037688e-06,
      "loss": 0.5146,
      "step": 1724
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.120188593864441,
      "learning_rate": 8.68567898067006e-06,
      "loss": 0.511,
      "step": 1725
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4266510009765625,
      "learning_rate": 8.68403465848286e-06,
      "loss": 0.4889,
      "step": 1726
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2310680150985718,
      "learning_rate": 8.682389464204526e-06,
      "loss": 0.4412,
      "step": 1727
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2980215549468994,
      "learning_rate": 8.680743398224511e-06,
      "loss": 0.5037,
      "step": 1728
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0948623418807983,
      "learning_rate": 8.679096460932477e-06,
      "loss": 0.4035,
      "step": 1729
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4676564931869507,
      "learning_rate": 8.677448652718286e-06,
      "loss": 0.5259,
      "step": 1730
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1023317575454712,
      "learning_rate": 8.675799973972012e-06,
      "loss": 0.4609,
      "step": 1731
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4761539697647095,
      "learning_rate": 8.67415042508393e-06,
      "loss": 0.5113,
      "step": 1732
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0517419576644897,
      "learning_rate": 8.67250000644453e-06,
      "loss": 0.4162,
      "step": 1733
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1588151454925537,
      "learning_rate": 8.6708487184445e-06,
      "loss": 0.4156,
      "step": 1734
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2297931909561157,
      "learning_rate": 8.669196561474735e-06,
      "loss": 0.4326,
      "step": 1735
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.460738182067871,
      "learning_rate": 8.667543535926335e-06,
      "loss": 0.5388,
      "step": 1736
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2446244955062866,
      "learning_rate": 8.665889642190608e-06,
      "loss": 0.5506,
      "step": 1737
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6825814247131348,
      "learning_rate": 8.664234880659068e-06,
      "loss": 0.4365,
      "step": 1738
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1229697465896606,
      "learning_rate": 8.662579251723431e-06,
      "loss": 0.4273,
      "step": 1739
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4379090070724487,
      "learning_rate": 8.660922755775622e-06,
      "loss": 0.5068,
      "step": 1740
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.632873058319092,
      "learning_rate": 8.659265393207769e-06,
      "loss": 0.4955,
      "step": 1741
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2496179342269897,
      "learning_rate": 8.657607164412208e-06,
      "loss": 0.5233,
      "step": 1742
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.437125563621521,
      "learning_rate": 8.655948069781475e-06,
      "loss": 0.5356,
      "step": 1743
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.797304391860962,
      "learning_rate": 8.654288109708314e-06,
      "loss": 0.4721,
      "step": 1744
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.500831961631775,
      "learning_rate": 8.652627284585674e-06,
      "loss": 0.4761,
      "step": 1745
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0811270475387573,
      "learning_rate": 8.650965594806707e-06,
      "loss": 0.4804,
      "step": 1746
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1000022888183594,
      "learning_rate": 8.649303040764774e-06,
      "loss": 0.4275,
      "step": 1747
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.033955693244934,
      "learning_rate": 8.647639622853436e-06,
      "loss": 0.4218,
      "step": 1748
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2953510284423828,
      "learning_rate": 8.645975341466462e-06,
      "loss": 0.4521,
      "step": 1749
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3148564100265503,
      "learning_rate": 8.64431019699782e-06,
      "loss": 0.4914,
      "step": 1750
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9009134769439697,
      "learning_rate": 8.64264418984169e-06,
      "loss": 0.44,
      "step": 1751
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2351820468902588,
      "learning_rate": 8.64097732039245e-06,
      "loss": 0.4647,
      "step": 1752
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0315431356430054,
      "learning_rate": 8.639309589044684e-06,
      "loss": 0.437,
      "step": 1753
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2227426767349243,
      "learning_rate": 8.637640996193179e-06,
      "loss": 0.4984,
      "step": 1754
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2556781768798828,
      "learning_rate": 8.635971542232933e-06,
      "loss": 0.5625,
      "step": 1755
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4592205286026,
      "learning_rate": 8.634301227559136e-06,
      "loss": 0.5107,
      "step": 1756
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.983941555023193,
      "learning_rate": 8.63263005256719e-06,
      "loss": 0.5223,
      "step": 1757
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.450717568397522,
      "learning_rate": 8.630958017652698e-06,
      "loss": 0.5033,
      "step": 1758
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6745820045471191,
      "learning_rate": 8.629285123211467e-06,
      "loss": 0.4691,
      "step": 1759
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.217221975326538,
      "learning_rate": 8.627611369639508e-06,
      "loss": 0.5139,
      "step": 1760
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.132144570350647,
      "learning_rate": 8.625936757333036e-06,
      "loss": 0.501,
      "step": 1761
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.418944001197815,
      "learning_rate": 8.624261286688467e-06,
      "loss": 0.4926,
      "step": 1762
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1488661766052246,
      "learning_rate": 8.62258495810242e-06,
      "loss": 0.4813,
      "step": 1763
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6462165117263794,
      "learning_rate": 8.62090777197172e-06,
      "loss": 0.5312,
      "step": 1764
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1584426164627075,
      "learning_rate": 8.619229728693393e-06,
      "loss": 0.4976,
      "step": 1765
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3042612075805664,
      "learning_rate": 8.617550828664669e-06,
      "loss": 0.4515,
      "step": 1766
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.274499535560608,
      "learning_rate": 8.61587107228298e-06,
      "loss": 0.5894,
      "step": 1767
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0410972833633423,
      "learning_rate": 8.614190459945959e-06,
      "loss": 0.4513,
      "step": 1768
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0396833419799805,
      "learning_rate": 8.612508992051447e-06,
      "loss": 0.5333,
      "step": 1769
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2684447765350342,
      "learning_rate": 8.61082666899748e-06,
      "loss": 0.5312,
      "step": 1770
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4789633750915527,
      "learning_rate": 8.609143491182303e-06,
      "loss": 0.5119,
      "step": 1771
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1795761585235596,
      "learning_rate": 8.607459459004362e-06,
      "loss": 0.443,
      "step": 1772
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4177404642105103,
      "learning_rate": 8.605774572862301e-06,
      "loss": 0.5647,
      "step": 1773
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.213029146194458,
      "learning_rate": 8.604088833154971e-06,
      "loss": 0.4978,
      "step": 1774
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6137021780014038,
      "learning_rate": 8.602402240281422e-06,
      "loss": 0.5124,
      "step": 1775
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1544246673583984,
      "learning_rate": 8.600714794640909e-06,
      "loss": 0.4895,
      "step": 1776
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.81827712059021,
      "learning_rate": 8.599026496632886e-06,
      "loss": 0.4584,
      "step": 1777
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0016533136367798,
      "learning_rate": 8.597337346657008e-06,
      "loss": 0.4075,
      "step": 1778
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5731699466705322,
      "learning_rate": 8.595647345113139e-06,
      "loss": 0.4575,
      "step": 1779
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.107572317123413,
      "learning_rate": 8.593956492401332e-06,
      "loss": 0.4518,
      "step": 1780
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3969473838806152,
      "learning_rate": 8.592264788921854e-06,
      "loss": 0.4581,
      "step": 1781
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3243895769119263,
      "learning_rate": 8.590572235075167e-06,
      "loss": 0.4762,
      "step": 1782
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5807334184646606,
      "learning_rate": 8.588878831261934e-06,
      "loss": 0.5994,
      "step": 1783
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6243295669555664,
      "learning_rate": 8.587184577883018e-06,
      "loss": 0.4444,
      "step": 1784
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2175236940383911,
      "learning_rate": 8.585489475339492e-06,
      "loss": 0.5566,
      "step": 1785
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1430132389068604,
      "learning_rate": 8.58379352403262e-06,
      "loss": 0.515,
      "step": 1786
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2192555665969849,
      "learning_rate": 8.58209672436387e-06,
      "loss": 0.5114,
      "step": 1787
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3097598552703857,
      "learning_rate": 8.580399076734913e-06,
      "loss": 0.4715,
      "step": 1788
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2860352993011475,
      "learning_rate": 8.578700581547619e-06,
      "loss": 0.5565,
      "step": 1789
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5543694496154785,
      "learning_rate": 8.577001239204056e-06,
      "loss": 0.5449,
      "step": 1790
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2634539604187012,
      "learning_rate": 8.5753010501065e-06,
      "loss": 0.7232,
      "step": 1791
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.216265082359314,
      "learning_rate": 8.57360001465742e-06,
      "loss": 0.5537,
      "step": 1792
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.369307518005371,
      "learning_rate": 8.571898133259487e-06,
      "loss": 0.4585,
      "step": 1793
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2553491592407227,
      "learning_rate": 8.570195406315575e-06,
      "loss": 0.4772,
      "step": 1794
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1978561878204346,
      "learning_rate": 8.568491834228757e-06,
      "loss": 0.5192,
      "step": 1795
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2588605880737305,
      "learning_rate": 8.566787417402302e-06,
      "loss": 0.5366,
      "step": 1796
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2389754056930542,
      "learning_rate": 8.565082156239689e-06,
      "loss": 0.7446,
      "step": 1797
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7997771501541138,
      "learning_rate": 8.563376051144582e-06,
      "loss": 0.4872,
      "step": 1798
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.486541509628296,
      "learning_rate": 8.561669102520861e-06,
      "loss": 0.5587,
      "step": 1799
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4997317790985107,
      "learning_rate": 8.559961310772595e-06,
      "loss": 0.5359,
      "step": 1800
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1134512424468994,
      "learning_rate": 8.558252676304052e-06,
      "loss": 0.5168,
      "step": 1801
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1551599502563477,
      "learning_rate": 8.55654319951971e-06,
      "loss": 0.4427,
      "step": 1802
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1063491106033325,
      "learning_rate": 8.554832880824233e-06,
      "loss": 0.574,
      "step": 1803
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.211456060409546,
      "learning_rate": 8.553121720622494e-06,
      "loss": 0.5094,
      "step": 1804
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.284637212753296,
      "learning_rate": 8.551409719319561e-06,
      "loss": 0.4457,
      "step": 1805
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4446903467178345,
      "learning_rate": 8.549696877320702e-06,
      "loss": 0.5112,
      "step": 1806
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1843383312225342,
      "learning_rate": 8.547983195031383e-06,
      "loss": 0.4818,
      "step": 1807
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6679799556732178,
      "learning_rate": 8.546268672857272e-06,
      "loss": 0.5168,
      "step": 1808
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4288721084594727,
      "learning_rate": 8.54455331120423e-06,
      "loss": 0.5061,
      "step": 1809
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.348368525505066,
      "learning_rate": 8.542837110478324e-06,
      "loss": 0.4356,
      "step": 1810
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1868518590927124,
      "learning_rate": 8.541120071085815e-06,
      "loss": 0.4718,
      "step": 1811
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2795665264129639,
      "learning_rate": 8.539402193433164e-06,
      "loss": 0.4766,
      "step": 1812
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3480616807937622,
      "learning_rate": 8.537683477927027e-06,
      "loss": 0.4395,
      "step": 1813
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1452692747116089,
      "learning_rate": 8.535963924974266e-06,
      "loss": 0.5689,
      "step": 1814
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1231529712677002,
      "learning_rate": 8.534243534981935e-06,
      "loss": 0.701,
      "step": 1815
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1872525215148926,
      "learning_rate": 8.532522308357285e-06,
      "loss": 0.485,
      "step": 1816
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3930408954620361,
      "learning_rate": 8.530800245507771e-06,
      "loss": 0.4616,
      "step": 1817
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3362256288528442,
      "learning_rate": 8.529077346841042e-06,
      "loss": 0.4903,
      "step": 1818
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.351198673248291,
      "learning_rate": 8.527353612764944e-06,
      "loss": 0.5313,
      "step": 1819
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2332661151885986,
      "learning_rate": 8.525629043687523e-06,
      "loss": 0.4787,
      "step": 1820
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1051915884017944,
      "learning_rate": 8.523903640017024e-06,
      "loss": 0.7099,
      "step": 1821
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4663296937942505,
      "learning_rate": 8.522177402161885e-06,
      "loss": 0.5669,
      "step": 1822
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2787610292434692,
      "learning_rate": 8.520450330530746e-06,
      "loss": 0.5156,
      "step": 1823
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3082901239395142,
      "learning_rate": 8.51872242553244e-06,
      "loss": 0.4841,
      "step": 1824
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3043811321258545,
      "learning_rate": 8.516993687576002e-06,
      "loss": 0.4999,
      "step": 1825
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3153692483901978,
      "learning_rate": 8.515264117070663e-06,
      "loss": 0.4798,
      "step": 1826
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.099561333656311,
      "learning_rate": 8.513533714425846e-06,
      "loss": 0.4679,
      "step": 1827
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2286393642425537,
      "learning_rate": 8.511802480051179e-06,
      "loss": 0.5016,
      "step": 1828
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.1166763305664062,
      "learning_rate": 8.510070414356478e-06,
      "loss": 0.4704,
      "step": 1829
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2182927131652832,
      "learning_rate": 8.508337517751765e-06,
      "loss": 0.5723,
      "step": 1830
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.587782859802246,
      "learning_rate": 8.506603790647252e-06,
      "loss": 0.4893,
      "step": 1831
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0778114795684814,
      "learning_rate": 8.504869233453352e-06,
      "loss": 0.4387,
      "step": 1832
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8863983154296875,
      "learning_rate": 8.503133846580671e-06,
      "loss": 0.576,
      "step": 1833
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.575267791748047,
      "learning_rate": 8.501397630440012e-06,
      "loss": 0.4498,
      "step": 1834
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5043951272964478,
      "learning_rate": 8.499660585442376e-06,
      "loss": 0.451,
      "step": 1835
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.402470111846924,
      "learning_rate": 8.49792271199896e-06,
      "loss": 0.4488,
      "step": 1836
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4568270444869995,
      "learning_rate": 8.496184010521155e-06,
      "loss": 0.4939,
      "step": 1837
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.357795238494873,
      "learning_rate": 8.494444481420552e-06,
      "loss": 0.4259,
      "step": 1838
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1875160932540894,
      "learning_rate": 8.492704125108933e-06,
      "loss": 0.4939,
      "step": 1839
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.151640772819519,
      "learning_rate": 8.490962941998278e-06,
      "loss": 0.4794,
      "step": 1840
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.2487411499023438,
      "learning_rate": 8.489220932500765e-06,
      "loss": 0.5086,
      "step": 1841
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4884222745895386,
      "learning_rate": 8.487478097028764e-06,
      "loss": 0.5121,
      "step": 1842
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.080124020576477,
      "learning_rate": 8.485734435994841e-06,
      "loss": 0.4957,
      "step": 1843
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1767922639846802,
      "learning_rate": 8.483989949811761e-06,
      "loss": 0.4744,
      "step": 1844
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0188641548156738,
      "learning_rate": 8.482244638892482e-06,
      "loss": 0.4638,
      "step": 1845
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3241477012634277,
      "learning_rate": 8.480498503650153e-06,
      "loss": 0.4484,
      "step": 1846
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.877872109413147,
      "learning_rate": 8.478751544498127e-06,
      "loss": 0.5003,
      "step": 1847
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1633023023605347,
      "learning_rate": 8.477003761849946e-06,
      "loss": 0.4437,
      "step": 1848
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.119922161102295,
      "learning_rate": 8.475255156119346e-06,
      "loss": 0.4902,
      "step": 1849
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2305899858474731,
      "learning_rate": 8.473505727720262e-06,
      "loss": 0.5074,
      "step": 1850
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.882646083831787,
      "learning_rate": 8.471755477066819e-06,
      "loss": 0.5203,
      "step": 1851
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4046772718429565,
      "learning_rate": 8.470004404573344e-06,
      "loss": 0.5019,
      "step": 1852
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0225887298583984,
      "learning_rate": 8.46825251065435e-06,
      "loss": 0.4766,
      "step": 1853
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.707108497619629,
      "learning_rate": 8.46649979572455e-06,
      "loss": 0.5021,
      "step": 1854
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6249297857284546,
      "learning_rate": 8.464746260198851e-06,
      "loss": 0.4789,
      "step": 1855
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3498066663742065,
      "learning_rate": 8.46299190449235e-06,
      "loss": 0.5147,
      "step": 1856
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2359771728515625,
      "learning_rate": 8.461236729020344e-06,
      "loss": 0.5777,
      "step": 1857
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7152513265609741,
      "learning_rate": 8.459480734198317e-06,
      "loss": 0.4888,
      "step": 1858
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2833305597305298,
      "learning_rate": 8.457723920441957e-06,
      "loss": 0.4498,
      "step": 1859
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2177934646606445,
      "learning_rate": 8.455966288167136e-06,
      "loss": 0.4358,
      "step": 1860
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.441273808479309,
      "learning_rate": 8.454207837789928e-06,
      "loss": 0.4837,
      "step": 1861
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3929473161697388,
      "learning_rate": 8.452448569726588e-06,
      "loss": 0.498,
      "step": 1862
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.357308030128479,
      "learning_rate": 8.450688484393583e-06,
      "loss": 0.5009,
      "step": 1863
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3253387212753296,
      "learning_rate": 8.448927582207556e-06,
      "loss": 0.4696,
      "step": 1864
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3199057579040527,
      "learning_rate": 8.447165863585355e-06,
      "loss": 0.483,
      "step": 1865
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0794177055358887,
      "learning_rate": 8.445403328944017e-06,
      "loss": 0.488,
      "step": 1866
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7082939147949219,
      "learning_rate": 8.44363997870077e-06,
      "loss": 0.4608,
      "step": 1867
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.33519446849823,
      "learning_rate": 8.441875813273038e-06,
      "loss": 0.4612,
      "step": 1868
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4753652811050415,
      "learning_rate": 8.440110833078438e-06,
      "loss": 0.4941,
      "step": 1869
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.249908208847046,
      "learning_rate": 8.43834503853478e-06,
      "loss": 0.5003,
      "step": 1870
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6126484870910645,
      "learning_rate": 8.436578430060064e-06,
      "loss": 0.458,
      "step": 1871
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3641928434371948,
      "learning_rate": 8.434811008072486e-06,
      "loss": 0.4504,
      "step": 1872
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2984997034072876,
      "learning_rate": 8.433042772990432e-06,
      "loss": 0.4847,
      "step": 1873
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3546440601348877,
      "learning_rate": 8.431273725232485e-06,
      "loss": 0.5196,
      "step": 1874
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3584585189819336,
      "learning_rate": 8.429503865217412e-06,
      "loss": 0.4816,
      "step": 1875
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.960636854171753,
      "learning_rate": 8.427733193364182e-06,
      "loss": 0.4794,
      "step": 1876
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8789170980453491,
      "learning_rate": 8.425961710091948e-06,
      "loss": 0.4302,
      "step": 1877
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7470546960830688,
      "learning_rate": 8.424189415820063e-06,
      "loss": 0.5273,
      "step": 1878
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5232231616973877,
      "learning_rate": 8.422416310968061e-06,
      "loss": 0.5559,
      "step": 1879
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4602501392364502,
      "learning_rate": 8.42064239595568e-06,
      "loss": 0.471,
      "step": 1880
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.346459150314331,
      "learning_rate": 8.418867671202844e-06,
      "loss": 0.4938,
      "step": 1881
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.978975296020508,
      "learning_rate": 8.417092137129665e-06,
      "loss": 0.5318,
      "step": 1882
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3658287525177002,
      "learning_rate": 8.415315794156456e-06,
      "loss": 0.5331,
      "step": 1883
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6069753170013428,
      "learning_rate": 8.413538642703708e-06,
      "loss": 0.4528,
      "step": 1884
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.322977900505066,
      "learning_rate": 8.41176068319212e-06,
      "loss": 0.4098,
      "step": 1885
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.502334713935852,
      "learning_rate": 8.409981916042572e-06,
      "loss": 0.5037,
      "step": 1886
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0371124744415283,
      "learning_rate": 8.408202341676132e-06,
      "loss": 0.4751,
      "step": 1887
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1848289966583252,
      "learning_rate": 8.406421960514068e-06,
      "loss": 0.5237,
      "step": 1888
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.534698724746704,
      "learning_rate": 8.404640772977834e-06,
      "loss": 0.4812,
      "step": 1889
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9245253801345825,
      "learning_rate": 8.402858779489075e-06,
      "loss": 0.4909,
      "step": 1890
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3297868967056274,
      "learning_rate": 8.40107598046963e-06,
      "loss": 0.6033,
      "step": 1891
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.53145432472229,
      "learning_rate": 8.399292376341523e-06,
      "loss": 0.5371,
      "step": 1892
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5890679359436035,
      "learning_rate": 8.397507967526975e-06,
      "loss": 0.4869,
      "step": 1893
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8994022607803345,
      "learning_rate": 8.395722754448392e-06,
      "loss": 0.5039,
      "step": 1894
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.451106309890747,
      "learning_rate": 8.393936737528375e-06,
      "loss": 0.5697,
      "step": 1895
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5530893802642822,
      "learning_rate": 8.39214991718971e-06,
      "loss": 0.4455,
      "step": 1896
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2611230611801147,
      "learning_rate": 8.390362293855378e-06,
      "loss": 0.4719,
      "step": 1897
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5102285146713257,
      "learning_rate": 8.388573867948548e-06,
      "loss": 0.475,
      "step": 1898
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3036892414093018,
      "learning_rate": 8.386784639892582e-06,
      "loss": 0.4827,
      "step": 1899
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2704391479492188,
      "learning_rate": 8.384994610111026e-06,
      "loss": 0.4551,
      "step": 1900
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3654756546020508,
      "learning_rate": 8.383203779027617e-06,
      "loss": 0.4968,
      "step": 1901
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1706453561782837,
      "learning_rate": 8.38141214706629e-06,
      "loss": 0.4625,
      "step": 1902
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1546274423599243,
      "learning_rate": 8.37961971465116e-06,
      "loss": 0.7047,
      "step": 1903
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1991535425186157,
      "learning_rate": 8.377826482206532e-06,
      "loss": 0.5143,
      "step": 1904
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2200731039047241,
      "learning_rate": 8.376032450156908e-06,
      "loss": 0.5031,
      "step": 1905
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1224360466003418,
      "learning_rate": 8.374237618926971e-06,
      "loss": 0.4468,
      "step": 1906
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5173063278198242,
      "learning_rate": 8.372441988941595e-06,
      "loss": 0.4735,
      "step": 1907
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.355027437210083,
      "learning_rate": 8.370645560625852e-06,
      "loss": 0.4806,
      "step": 1908
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9432934522628784,
      "learning_rate": 8.368848334404987e-06,
      "loss": 0.5138,
      "step": 1909
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3672922849655151,
      "learning_rate": 8.367050310704448e-06,
      "loss": 0.5412,
      "step": 1910
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6080561876296997,
      "learning_rate": 8.365251489949866e-06,
      "loss": 0.5454,
      "step": 1911
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.507300853729248,
      "learning_rate": 8.363451872567057e-06,
      "loss": 0.4963,
      "step": 1912
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4726314544677734,
      "learning_rate": 8.361651458982034e-06,
      "loss": 0.5626,
      "step": 1913
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.252935767173767,
      "learning_rate": 8.359850249620993e-06,
      "loss": 0.4321,
      "step": 1914
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0557688474655151,
      "learning_rate": 8.358048244910319e-06,
      "loss": 0.4346,
      "step": 1915
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1656901836395264,
      "learning_rate": 8.356245445276585e-06,
      "loss": 0.4366,
      "step": 1916
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.464011311531067,
      "learning_rate": 8.354441851146552e-06,
      "loss": 0.5091,
      "step": 1917
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1307380199432373,
      "learning_rate": 8.352637462947173e-06,
      "loss": 0.5027,
      "step": 1918
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4712284803390503,
      "learning_rate": 8.350832281105584e-06,
      "loss": 0.5286,
      "step": 1919
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1306713819503784,
      "learning_rate": 8.349026306049113e-06,
      "loss": 0.3883,
      "step": 1920
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2439597845077515,
      "learning_rate": 8.347219538205269e-06,
      "loss": 0.5255,
      "step": 1921
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.608749508857727,
      "learning_rate": 8.345411978001757e-06,
      "loss": 0.4474,
      "step": 1922
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.313237190246582,
      "learning_rate": 8.343603625866464e-06,
      "loss": 0.41,
      "step": 1923
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5865063667297363,
      "learning_rate": 8.341794482227467e-06,
      "loss": 0.4538,
      "step": 1924
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.244460940361023,
      "learning_rate": 8.339984547513031e-06,
      "loss": 0.5068,
      "step": 1925
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3845325708389282,
      "learning_rate": 8.338173822151607e-06,
      "loss": 0.4134,
      "step": 1926
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1595313549041748,
      "learning_rate": 8.33636230657183e-06,
      "loss": 0.4791,
      "step": 1927
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2204166650772095,
      "learning_rate": 8.334550001202526e-06,
      "loss": 0.4949,
      "step": 1928
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7885699272155762,
      "learning_rate": 8.33273690647271e-06,
      "loss": 0.4809,
      "step": 1929
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5315159559249878,
      "learning_rate": 8.330923022811576e-06,
      "loss": 0.4969,
      "step": 1930
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5264558792114258,
      "learning_rate": 8.329108350648516e-06,
      "loss": 0.5141,
      "step": 1931
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2993558645248413,
      "learning_rate": 8.327292890413096e-06,
      "loss": 0.5194,
      "step": 1932
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2263864278793335,
      "learning_rate": 8.32547664253508e-06,
      "loss": 0.7627,
      "step": 1933
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5256565809249878,
      "learning_rate": 8.32365960744441e-06,
      "loss": 0.4813,
      "step": 1934
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.250393033027649,
      "learning_rate": 8.321841785571221e-06,
      "loss": 0.491,
      "step": 1935
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.818467378616333,
      "learning_rate": 8.320023177345827e-06,
      "loss": 0.4423,
      "step": 1936
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.157051920890808,
      "learning_rate": 8.318203783198734e-06,
      "loss": 0.4915,
      "step": 1937
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.270466923713684,
      "learning_rate": 8.316383603560634e-06,
      "loss": 0.4662,
      "step": 1938
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6075917482376099,
      "learning_rate": 8.314562638862401e-06,
      "loss": 0.5038,
      "step": 1939
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3481591939926147,
      "learning_rate": 8.312740889535096e-06,
      "loss": 0.5128,
      "step": 1940
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3743871450424194,
      "learning_rate": 8.31091835600997e-06,
      "loss": 0.5097,
      "step": 1941
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.87348210811615,
      "learning_rate": 8.309095038718453e-06,
      "loss": 0.4698,
      "step": 1942
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2874494791030884,
      "learning_rate": 8.307270938092166e-06,
      "loss": 0.4183,
      "step": 1943
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.353935956954956,
      "learning_rate": 8.305446054562912e-06,
      "loss": 0.503,
      "step": 1944
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4547042846679688,
      "learning_rate": 8.303620388562681e-06,
      "loss": 0.5102,
      "step": 1945
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.2274422645568848,
      "learning_rate": 8.301793940523648e-06,
      "loss": 0.4864,
      "step": 1946
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6371383666992188,
      "learning_rate": 8.299966710878173e-06,
      "loss": 0.5171,
      "step": 1947
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6548800468444824,
      "learning_rate": 8.298138700058802e-06,
      "loss": 0.549,
      "step": 1948
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2856045961380005,
      "learning_rate": 8.296309908498264e-06,
      "loss": 0.455,
      "step": 1949
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5280119180679321,
      "learning_rate": 8.294480336629474e-06,
      "loss": 0.46,
      "step": 1950
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5763583183288574,
      "learning_rate": 8.292649984885528e-06,
      "loss": 0.4496,
      "step": 1951
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6710022687911987,
      "learning_rate": 8.290818853699716e-06,
      "loss": 0.4108,
      "step": 1952
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0640249252319336,
      "learning_rate": 8.288986943505502e-06,
      "loss": 0.4592,
      "step": 1953
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7375167608261108,
      "learning_rate": 8.287154254736543e-06,
      "loss": 0.5021,
      "step": 1954
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3885396718978882,
      "learning_rate": 8.285320787826672e-06,
      "loss": 0.4473,
      "step": 1955
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2883998155593872,
      "learning_rate": 8.283486543209913e-06,
      "loss": 0.5156,
      "step": 1956
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4163599014282227,
      "learning_rate": 8.281651521320471e-06,
      "loss": 0.4592,
      "step": 1957
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5688552856445312,
      "learning_rate": 8.279815722592738e-06,
      "loss": 0.5083,
      "step": 1958
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4987449645996094,
      "learning_rate": 8.27797914746128e-06,
      "loss": 0.4397,
      "step": 1959
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4540972709655762,
      "learning_rate": 8.276141796360866e-06,
      "loss": 0.4891,
      "step": 1960
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2465580701828003,
      "learning_rate": 8.274303669726427e-06,
      "loss": 0.457,
      "step": 1961
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6686251163482666,
      "learning_rate": 8.27246476799309e-06,
      "loss": 0.4742,
      "step": 1962
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7289750576019287,
      "learning_rate": 8.270625091596164e-06,
      "loss": 0.5045,
      "step": 1963
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6876245737075806,
      "learning_rate": 8.268784640971143e-06,
      "loss": 0.5557,
      "step": 1964
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4175456762313843,
      "learning_rate": 8.266943416553698e-06,
      "loss": 0.459,
      "step": 1965
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.562124252319336,
      "learning_rate": 8.265101418779688e-06,
      "loss": 0.5157,
      "step": 1966
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7185218334197998,
      "learning_rate": 8.263258648085155e-06,
      "loss": 0.3937,
      "step": 1967
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1874008178710938,
      "learning_rate": 8.261415104906321e-06,
      "loss": 0.5414,
      "step": 1968
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5357635021209717,
      "learning_rate": 8.259570789679593e-06,
      "loss": 0.5014,
      "step": 1969
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6194719076156616,
      "learning_rate": 8.257725702841562e-06,
      "loss": 0.474,
      "step": 1970
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3717339038848877,
      "learning_rate": 8.255879844829e-06,
      "loss": 0.4274,
      "step": 1971
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.555689811706543,
      "learning_rate": 8.25403321607886e-06,
      "loss": 0.5211,
      "step": 1972
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.905397891998291,
      "learning_rate": 8.25218581702828e-06,
      "loss": 0.5003,
      "step": 1973
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5162171125411987,
      "learning_rate": 8.25033764811458e-06,
      "loss": 0.6049,
      "step": 1974
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8621119260787964,
      "learning_rate": 8.248488709775262e-06,
      "loss": 0.4766,
      "step": 1975
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4944742918014526,
      "learning_rate": 8.24663900244801e-06,
      "loss": 0.4835,
      "step": 1976
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.398222804069519,
      "learning_rate": 8.24478852657069e-06,
      "loss": 0.4882,
      "step": 1977
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4713521003723145,
      "learning_rate": 8.242937282581348e-06,
      "loss": 0.5114,
      "step": 1978
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7327511310577393,
      "learning_rate": 8.241085270918215e-06,
      "loss": 0.4482,
      "step": 1979
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.453220844268799,
      "learning_rate": 8.239232492019702e-06,
      "loss": 0.4625,
      "step": 1980
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.817134141921997,
      "learning_rate": 8.237378946324404e-06,
      "loss": 0.358,
      "step": 1981
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5401021242141724,
      "learning_rate": 8.235524634271095e-06,
      "loss": 0.5654,
      "step": 1982
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4683477878570557,
      "learning_rate": 8.23366955629873e-06,
      "loss": 0.5617,
      "step": 1983
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5621545314788818,
      "learning_rate": 8.231813712846445e-06,
      "loss": 0.439,
      "step": 1984
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.300576090812683,
      "learning_rate": 8.22995710435356e-06,
      "loss": 0.4688,
      "step": 1985
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.019500255584717,
      "learning_rate": 8.228099731259575e-06,
      "loss": 0.4257,
      "step": 1986
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4027212858200073,
      "learning_rate": 8.22624159400417e-06,
      "loss": 0.4385,
      "step": 1987
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3703433275222778,
      "learning_rate": 8.224382693027207e-06,
      "loss": 0.4984,
      "step": 1988
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2030946016311646,
      "learning_rate": 8.22252302876873e-06,
      "loss": 0.7422,
      "step": 1989
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3134950399398804,
      "learning_rate": 8.220662601668956e-06,
      "loss": 0.7761,
      "step": 1990
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6618233919143677,
      "learning_rate": 8.218801412168297e-06,
      "loss": 0.4703,
      "step": 1991
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6034631729125977,
      "learning_rate": 8.21693946070733e-06,
      "loss": 0.456,
      "step": 1992
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7630822658538818,
      "learning_rate": 8.215076747726821e-06,
      "loss": 0.5332,
      "step": 1993
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.1593146324157715,
      "learning_rate": 8.213213273667718e-06,
      "loss": 0.5191,
      "step": 1994
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8803586959838867,
      "learning_rate": 8.211349038971142e-06,
      "loss": 0.5181,
      "step": 1995
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.83259916305542,
      "learning_rate": 8.2094840440784e-06,
      "loss": 0.4858,
      "step": 1996
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0749820470809937,
      "learning_rate": 8.207618289430975e-06,
      "loss": 0.7257,
      "step": 1997
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7103792428970337,
      "learning_rate": 8.205751775470535e-06,
      "loss": 0.5176,
      "step": 1998
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.1109890937805176,
      "learning_rate": 8.203884502638921e-06,
      "loss": 0.5075,
      "step": 1999
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.538441777229309,
      "learning_rate": 8.202016471378161e-06,
      "loss": 0.5309,
      "step": 2000
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.19952392578125,
      "learning_rate": 8.200147682130452e-06,
      "loss": 0.489,
      "step": 2001
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9257923364639282,
      "learning_rate": 8.198278135338185e-06,
      "loss": 0.4185,
      "step": 2002
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1837838888168335,
      "learning_rate": 8.196407831443917e-06,
      "loss": 0.7515,
      "step": 2003
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.04306960105896,
      "learning_rate": 8.194536770890392e-06,
      "loss": 0.5209,
      "step": 2004
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2554807662963867,
      "learning_rate": 8.192664954120528e-06,
      "loss": 0.7702,
      "step": 2005
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8789118528366089,
      "learning_rate": 8.190792381577429e-06,
      "loss": 0.5202,
      "step": 2006
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.1097216606140137,
      "learning_rate": 8.18891905370437e-06,
      "loss": 0.4293,
      "step": 2007
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.974115014076233,
      "learning_rate": 8.187044970944811e-06,
      "loss": 0.4735,
      "step": 2008
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6766105890274048,
      "learning_rate": 8.185170133742386e-06,
      "loss": 0.5518,
      "step": 2009
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.7787842750549316,
      "learning_rate": 8.183294542540912e-06,
      "loss": 0.5135,
      "step": 2010
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8457355499267578,
      "learning_rate": 8.181418197784379e-06,
      "loss": 0.5023,
      "step": 2011
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7436408996582031,
      "learning_rate": 8.179541099916961e-06,
      "loss": 0.4863,
      "step": 2012
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7818166017532349,
      "learning_rate": 8.177663249383008e-06,
      "loss": 0.4822,
      "step": 2013
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7465578317642212,
      "learning_rate": 8.175784646627048e-06,
      "loss": 0.4765,
      "step": 2014
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.1486454010009766,
      "learning_rate": 8.173905292093784e-06,
      "loss": 0.5183,
      "step": 2015
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8642292022705078,
      "learning_rate": 8.172025186228107e-06,
      "loss": 0.3994,
      "step": 2016
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.235391616821289,
      "learning_rate": 8.170144329475071e-06,
      "loss": 0.422,
      "step": 2017
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.537213087081909,
      "learning_rate": 8.168262722279921e-06,
      "loss": 0.4445,
      "step": 2018
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.3780770301818848,
      "learning_rate": 8.166380365088074e-06,
      "loss": 0.4933,
      "step": 2019
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.924559235572815,
      "learning_rate": 8.164497258345121e-06,
      "loss": 0.425,
      "step": 2020
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.3788082599639893,
      "learning_rate": 8.162613402496839e-06,
      "loss": 0.4506,
      "step": 2021
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.452616572380066,
      "learning_rate": 8.160728797989175e-06,
      "loss": 0.4671,
      "step": 2022
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.1409382820129395,
      "learning_rate": 8.158843445268256e-06,
      "loss": 0.4524,
      "step": 2023
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.9938368797302246,
      "learning_rate": 8.156957344780386e-06,
      "loss": 0.4715,
      "step": 2024
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.5160183906555176,
      "learning_rate": 8.155070496972049e-06,
      "loss": 0.4566,
      "step": 2025
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.2666497230529785,
      "learning_rate": 8.153182902289898e-06,
      "loss": 0.5416,
      "step": 2026
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.032745599746704,
      "learning_rate": 8.15129456118077e-06,
      "loss": 0.454,
      "step": 2027
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.9161441326141357,
      "learning_rate": 8.149405474091678e-06,
      "loss": 0.4682,
      "step": 2028
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.9612815380096436,
      "learning_rate": 8.147515641469808e-06,
      "loss": 0.523,
      "step": 2029
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9864650964736938,
      "learning_rate": 8.145625063762525e-06,
      "loss": 0.4388,
      "step": 2030
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.243927001953125,
      "learning_rate": 8.14373374141737e-06,
      "loss": 0.4242,
      "step": 2031
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9632470607757568,
      "learning_rate": 8.14184167488206e-06,
      "loss": 0.5263,
      "step": 2032
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.16975474357605,
      "learning_rate": 8.139948864604488e-06,
      "loss": 0.5267,
      "step": 2033
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6179780960083008,
      "learning_rate": 8.138055311032724e-06,
      "loss": 0.6014,
      "step": 2034
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6028002500534058,
      "learning_rate": 8.136161014615013e-06,
      "loss": 0.4999,
      "step": 2035
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.574171781539917,
      "learning_rate": 8.134265975799775e-06,
      "loss": 0.4583,
      "step": 2036
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8450121879577637,
      "learning_rate": 8.13237019503561e-06,
      "loss": 0.4446,
      "step": 2037
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5149773359298706,
      "learning_rate": 8.13047367277129e-06,
      "loss": 0.4424,
      "step": 2038
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9277148246765137,
      "learning_rate": 8.128576409455759e-06,
      "loss": 0.5219,
      "step": 2039
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.600396990776062,
      "learning_rate": 8.126678405538143e-06,
      "loss": 0.493,
      "step": 2040
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6990512609481812,
      "learning_rate": 8.12477966146774e-06,
      "loss": 0.4196,
      "step": 2041
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.026313304901123,
      "learning_rate": 8.122880177694028e-06,
      "loss": 0.4783,
      "step": 2042
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.428992748260498,
      "learning_rate": 8.120979954666652e-06,
      "loss": 0.4733,
      "step": 2043
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.3387451171875,
      "learning_rate": 8.119078992835439e-06,
      "loss": 0.4679,
      "step": 2044
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5320236682891846,
      "learning_rate": 8.117177292650384e-06,
      "loss": 0.4754,
      "step": 2045
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8779270648956299,
      "learning_rate": 8.115274854561663e-06,
      "loss": 0.5164,
      "step": 2046
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.319556951522827,
      "learning_rate": 8.113371679019625e-06,
      "loss": 0.4903,
      "step": 2047
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3922730684280396,
      "learning_rate": 8.111467766474793e-06,
      "loss": 0.4668,
      "step": 2048
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.029789924621582,
      "learning_rate": 8.109563117377865e-06,
      "loss": 0.4247,
      "step": 2049
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.3358991146087646,
      "learning_rate": 8.10765773217971e-06,
      "loss": 0.4338,
      "step": 2050
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6662729978561401,
      "learning_rate": 8.105751611331377e-06,
      "loss": 0.5031,
      "step": 2051
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.844216227531433,
      "learning_rate": 8.103844755284086e-06,
      "loss": 0.505,
      "step": 2052
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.420297145843506,
      "learning_rate": 8.10193716448923e-06,
      "loss": 0.4957,
      "step": 2053
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9902347326278687,
      "learning_rate": 8.100028839398377e-06,
      "loss": 0.4546,
      "step": 2054
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.546780824661255,
      "learning_rate": 8.098119780463271e-06,
      "loss": 0.4638,
      "step": 2055
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4772707223892212,
      "learning_rate": 8.096209988135828e-06,
      "loss": 0.4976,
      "step": 2056
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4665801525115967,
      "learning_rate": 8.094299462868134e-06,
      "loss": 0.4245,
      "step": 2057
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3346469402313232,
      "learning_rate": 8.092388205112455e-06,
      "loss": 0.4313,
      "step": 2058
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7558096647262573,
      "learning_rate": 8.090476215321226e-06,
      "loss": 0.5137,
      "step": 2059
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4338704347610474,
      "learning_rate": 8.088563493947058e-06,
      "loss": 0.4824,
      "step": 2060
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.235007405281067,
      "learning_rate": 8.08665004144273e-06,
      "loss": 0.7473,
      "step": 2061
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.170661687850952,
      "learning_rate": 8.0847358582612e-06,
      "loss": 0.4914,
      "step": 2062
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9872759580612183,
      "learning_rate": 8.082820944855601e-06,
      "loss": 0.5361,
      "step": 2063
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.363355875015259,
      "learning_rate": 8.080905301679229e-06,
      "loss": 0.4652,
      "step": 2064
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6600914001464844,
      "learning_rate": 8.07898892918556e-06,
      "loss": 0.5319,
      "step": 2065
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6764663457870483,
      "learning_rate": 8.077071827828242e-06,
      "loss": 0.4657,
      "step": 2066
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.712711811065674,
      "learning_rate": 8.075153998061094e-06,
      "loss": 0.4598,
      "step": 2067
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0049610137939453,
      "learning_rate": 8.073235440338107e-06,
      "loss": 0.4637,
      "step": 2068
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9091159105300903,
      "learning_rate": 8.071316155113446e-06,
      "loss": 0.4689,
      "step": 2069
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6682074069976807,
      "learning_rate": 8.069396142841451e-06,
      "loss": 0.4914,
      "step": 2070
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7776962518692017,
      "learning_rate": 8.067475403976625e-06,
      "loss": 0.4957,
      "step": 2071
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.631505012512207,
      "learning_rate": 8.065553938973652e-06,
      "loss": 0.438,
      "step": 2072
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7370164394378662,
      "learning_rate": 8.063631748287381e-06,
      "loss": 0.5375,
      "step": 2073
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7474329471588135,
      "learning_rate": 8.061708832372841e-06,
      "loss": 0.4235,
      "step": 2074
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6390455961227417,
      "learning_rate": 8.059785191685227e-06,
      "loss": 0.4176,
      "step": 2075
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.805836796760559,
      "learning_rate": 8.057860826679906e-06,
      "loss": 0.4868,
      "step": 2076
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3713089227676392,
      "learning_rate": 8.055935737812415e-06,
      "loss": 0.5055,
      "step": 2077
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.428264856338501,
      "learning_rate": 8.054009925538467e-06,
      "loss": 0.5113,
      "step": 2078
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.326550006866455,
      "learning_rate": 8.052083390313943e-06,
      "loss": 0.4708,
      "step": 2079
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.223088502883911,
      "learning_rate": 8.050156132594896e-06,
      "loss": 0.5155,
      "step": 2080
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.175442099571228,
      "learning_rate": 8.048228152837548e-06,
      "loss": 0.6719,
      "step": 2081
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9265012741088867,
      "learning_rate": 8.046299451498297e-06,
      "loss": 0.5574,
      "step": 2082
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.492128610610962,
      "learning_rate": 8.044370029033706e-06,
      "loss": 0.5202,
      "step": 2083
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8300371170043945,
      "learning_rate": 8.04243988590051e-06,
      "loss": 0.4162,
      "step": 2084
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6425495147705078,
      "learning_rate": 8.04050902255562e-06,
      "loss": 0.5294,
      "step": 2085
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.018345594406128,
      "learning_rate": 8.03857743945611e-06,
      "loss": 0.5231,
      "step": 2086
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5163805484771729,
      "learning_rate": 8.03664513705923e-06,
      "loss": 0.512,
      "step": 2087
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3836709260940552,
      "learning_rate": 8.034712115822396e-06,
      "loss": 0.5097,
      "step": 2088
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.5436294078826904,
      "learning_rate": 8.032778376203199e-06,
      "loss": 0.4328,
      "step": 2089
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6127201318740845,
      "learning_rate": 8.030843918659394e-06,
      "loss": 0.4973,
      "step": 2090
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.065004825592041,
      "learning_rate": 8.02890874364891e-06,
      "loss": 0.4581,
      "step": 2091
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7143522500991821,
      "learning_rate": 8.026972851629846e-06,
      "loss": 0.4875,
      "step": 2092
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.2647974491119385,
      "learning_rate": 8.025036243060469e-06,
      "loss": 0.4856,
      "step": 2093
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.349121332168579,
      "learning_rate": 8.023098918399218e-06,
      "loss": 0.5127,
      "step": 2094
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.952348232269287,
      "learning_rate": 8.021160878104697e-06,
      "loss": 0.5247,
      "step": 2095
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8046070337295532,
      "learning_rate": 8.019222122635684e-06,
      "loss": 0.4842,
      "step": 2096
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2088450193405151,
      "learning_rate": 8.017282652451127e-06,
      "loss": 0.7255,
      "step": 2097
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.314833164215088,
      "learning_rate": 8.015342468010134e-06,
      "loss": 0.488,
      "step": 2098
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1092438697814941,
      "learning_rate": 8.013401569771995e-06,
      "loss": 0.7175,
      "step": 2099
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.310930848121643,
      "learning_rate": 8.011459958196161e-06,
      "loss": 0.5107,
      "step": 2100
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.862406611442566,
      "learning_rate": 8.009517633742255e-06,
      "loss": 0.4265,
      "step": 2101
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.448464035987854,
      "learning_rate": 8.007574596870062e-06,
      "loss": 0.5173,
      "step": 2102
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8831887245178223,
      "learning_rate": 8.00563084803955e-06,
      "loss": 0.4861,
      "step": 2103
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.754654049873352,
      "learning_rate": 8.003686387710839e-06,
      "loss": 0.4198,
      "step": 2104
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.566817045211792,
      "learning_rate": 8.001741216344228e-06,
      "loss": 0.4423,
      "step": 2105
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.178847551345825,
      "learning_rate": 7.999795334400182e-06,
      "loss": 0.4628,
      "step": 2106
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4509711265563965,
      "learning_rate": 7.99784874233933e-06,
      "loss": 0.5213,
      "step": 2107
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3924716711044312,
      "learning_rate": 7.995901440622477e-06,
      "loss": 0.4666,
      "step": 2108
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7722052335739136,
      "learning_rate": 7.993953429710593e-06,
      "loss": 0.5422,
      "step": 2109
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8525986671447754,
      "learning_rate": 7.99200471006481e-06,
      "loss": 0.5157,
      "step": 2110
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3005156517028809,
      "learning_rate": 7.990055282146431e-06,
      "loss": 0.4506,
      "step": 2111
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7269099950790405,
      "learning_rate": 7.988105146416934e-06,
      "loss": 0.4978,
      "step": 2112
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6103154420852661,
      "learning_rate": 7.986154303337954e-06,
      "loss": 0.4631,
      "step": 2113
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2780239582061768,
      "learning_rate": 7.9842027533713e-06,
      "loss": 0.43,
      "step": 2114
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6896605491638184,
      "learning_rate": 7.982250496978948e-06,
      "loss": 0.4984,
      "step": 2115
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.118579864501953,
      "learning_rate": 7.980297534623034e-06,
      "loss": 0.4763,
      "step": 2116
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.597187876701355,
      "learning_rate": 7.978343866765871e-06,
      "loss": 0.4537,
      "step": 2117
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4828904867172241,
      "learning_rate": 7.976389493869934e-06,
      "loss": 0.5306,
      "step": 2118
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.347874402999878,
      "learning_rate": 7.974434416397866e-06,
      "loss": 0.4955,
      "step": 2119
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7554595470428467,
      "learning_rate": 7.972478634812475e-06,
      "loss": 0.5014,
      "step": 2120
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3867886066436768,
      "learning_rate": 7.970522149576738e-06,
      "loss": 0.4202,
      "step": 2121
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8210101127624512,
      "learning_rate": 7.968564961153796e-06,
      "loss": 0.4241,
      "step": 2122
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.546552300453186,
      "learning_rate": 7.966607070006961e-06,
      "loss": 0.4911,
      "step": 2123
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4727166891098022,
      "learning_rate": 7.964648476599706e-06,
      "loss": 0.4181,
      "step": 2124
    },
    {
      "epoch": 0.32,
      "grad_norm": 11.545703887939453,
      "learning_rate": 7.962689181395672e-06,
      "loss": 0.4657,
      "step": 2125
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7349592447280884,
      "learning_rate": 7.96072918485867e-06,
      "loss": 0.4828,
      "step": 2126
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2737489938735962,
      "learning_rate": 7.958768487452674e-06,
      "loss": 0.4403,
      "step": 2127
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8456597328186035,
      "learning_rate": 7.95680708964182e-06,
      "loss": 0.4749,
      "step": 2128
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.606020450592041,
      "learning_rate": 7.954844991890414e-06,
      "loss": 0.4159,
      "step": 2129
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4031306505203247,
      "learning_rate": 7.95288219466293e-06,
      "loss": 0.4031,
      "step": 2130
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.233870029449463,
      "learning_rate": 7.950918698424001e-06,
      "loss": 0.5316,
      "step": 2131
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4270081520080566,
      "learning_rate": 7.948954503638434e-06,
      "loss": 0.5193,
      "step": 2132
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4008878469467163,
      "learning_rate": 7.946989610771194e-06,
      "loss": 0.5041,
      "step": 2133
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5428613424301147,
      "learning_rate": 7.945024020287415e-06,
      "loss": 0.4624,
      "step": 2134
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6394795179367065,
      "learning_rate": 7.94305773265239e-06,
      "loss": 0.3981,
      "step": 2135
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4271423816680908,
      "learning_rate": 7.94109074833159e-06,
      "loss": 0.4629,
      "step": 2136
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4991190433502197,
      "learning_rate": 7.939123067790638e-06,
      "loss": 0.5513,
      "step": 2137
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9883553981781006,
      "learning_rate": 7.937154691495328e-06,
      "loss": 0.4497,
      "step": 2138
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9642915725708008,
      "learning_rate": 7.935185619911619e-06,
      "loss": 0.4574,
      "step": 2139
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7682828903198242,
      "learning_rate": 7.933215853505627e-06,
      "loss": 0.4883,
      "step": 2140
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5106993913650513,
      "learning_rate": 7.931245392743647e-06,
      "loss": 0.4739,
      "step": 2141
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5122348070144653,
      "learning_rate": 7.929274238092125e-06,
      "loss": 0.5475,
      "step": 2142
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.751315951347351,
      "learning_rate": 7.927302390017677e-06,
      "loss": 0.5324,
      "step": 2143
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.505202293395996,
      "learning_rate": 7.925329848987082e-06,
      "loss": 0.5006,
      "step": 2144
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.180228352546692,
      "learning_rate": 7.923356615467284e-06,
      "loss": 0.723,
      "step": 2145
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9550150632858276,
      "learning_rate": 7.92138268992539e-06,
      "loss": 0.4155,
      "step": 2146
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.42575204372406,
      "learning_rate": 7.919408072828671e-06,
      "loss": 0.5246,
      "step": 2147
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6339794397354126,
      "learning_rate": 7.917432764644564e-06,
      "loss": 0.4073,
      "step": 2148
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.362960934638977,
      "learning_rate": 7.915456765840664e-06,
      "loss": 0.4293,
      "step": 2149
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.407341718673706,
      "learning_rate": 7.913480076884733e-06,
      "loss": 0.4415,
      "step": 2150
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4733400344848633,
      "learning_rate": 7.9115026982447e-06,
      "loss": 0.4904,
      "step": 2151
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4333195686340332,
      "learning_rate": 7.90952463038865e-06,
      "loss": 0.4386,
      "step": 2152
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3839186429977417,
      "learning_rate": 7.907545873784836e-06,
      "loss": 0.4803,
      "step": 2153
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7453258037567139,
      "learning_rate": 7.905566428901672e-06,
      "loss": 0.4792,
      "step": 2154
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.7215559482574463,
      "learning_rate": 7.903586296207737e-06,
      "loss": 0.521,
      "step": 2155
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4303606748580933,
      "learning_rate": 7.90160547617177e-06,
      "loss": 0.4889,
      "step": 2156
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.395446538925171,
      "learning_rate": 7.899623969262676e-06,
      "loss": 0.4743,
      "step": 2157
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0869030952453613,
      "learning_rate": 7.897641775949518e-06,
      "loss": 0.4401,
      "step": 2158
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.9039146900177,
      "learning_rate": 7.895658896701526e-06,
      "loss": 0.5022,
      "step": 2159
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.479784607887268,
      "learning_rate": 7.893675331988093e-06,
      "loss": 0.4417,
      "step": 2160
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2903202772140503,
      "learning_rate": 7.891691082278768e-06,
      "loss": 0.3982,
      "step": 2161
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.707764744758606,
      "learning_rate": 7.889706148043266e-06,
      "loss": 0.4274,
      "step": 2162
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3869061470031738,
      "learning_rate": 7.887720529751467e-06,
      "loss": 0.5101,
      "step": 2163
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.222954750061035,
      "learning_rate": 7.885734227873409e-06,
      "loss": 0.3718,
      "step": 2164
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0343117713928223,
      "learning_rate": 7.883747242879292e-06,
      "loss": 0.4085,
      "step": 2165
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1687328815460205,
      "learning_rate": 7.881759575239481e-06,
      "loss": 0.6934,
      "step": 2166
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2004035711288452,
      "learning_rate": 7.879771225424496e-06,
      "loss": 0.4319,
      "step": 2167
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7771553993225098,
      "learning_rate": 7.877782193905025e-06,
      "loss": 0.3423,
      "step": 2168
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5419038534164429,
      "learning_rate": 7.875792481151916e-06,
      "loss": 0.4639,
      "step": 2169
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.147688388824463,
      "learning_rate": 7.873802087636175e-06,
      "loss": 0.537,
      "step": 2170
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0654979944229126,
      "learning_rate": 7.871811013828973e-06,
      "loss": 0.4703,
      "step": 2171
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2188739776611328,
      "learning_rate": 7.869819260201643e-06,
      "loss": 0.4927,
      "step": 2172
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9836480617523193,
      "learning_rate": 7.86782682722567e-06,
      "loss": 0.4649,
      "step": 2173
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3067309856414795,
      "learning_rate": 7.865833715372711e-06,
      "loss": 0.4342,
      "step": 2174
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4352444410324097,
      "learning_rate": 7.863839925114578e-06,
      "loss": 0.505,
      "step": 2175
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7680106163024902,
      "learning_rate": 7.861845456923244e-06,
      "loss": 0.4627,
      "step": 2176
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.467600107192993,
      "learning_rate": 7.859850311270844e-06,
      "loss": 0.5033,
      "step": 2177
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8287923336029053,
      "learning_rate": 7.85785448862967e-06,
      "loss": 0.4771,
      "step": 2178
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.499864935874939,
      "learning_rate": 7.85585798947218e-06,
      "loss": 0.4449,
      "step": 2179
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.38177490234375,
      "learning_rate": 7.853860814270987e-06,
      "loss": 0.4805,
      "step": 2180
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1276488304138184,
      "learning_rate": 7.851862963498864e-06,
      "loss": 0.728,
      "step": 2181
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1927435398101807,
      "learning_rate": 7.84986443762875e-06,
      "loss": 0.4939,
      "step": 2182
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4852851629257202,
      "learning_rate": 7.847865237133735e-06,
      "loss": 0.4336,
      "step": 2183
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.333211064338684,
      "learning_rate": 7.845865362487077e-06,
      "loss": 0.4968,
      "step": 2184
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.3353030681610107,
      "learning_rate": 7.843864814162185e-06,
      "loss": 0.4574,
      "step": 2185
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.327622890472412,
      "learning_rate": 7.84186359263264e-06,
      "loss": 0.4351,
      "step": 2186
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7811737060546875,
      "learning_rate": 7.839861698372166e-06,
      "loss": 0.4448,
      "step": 2187
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8070697784423828,
      "learning_rate": 7.837859131854662e-06,
      "loss": 0.4655,
      "step": 2188
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2274051904678345,
      "learning_rate": 7.835855893554174e-06,
      "loss": 0.521,
      "step": 2189
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0660635232925415,
      "learning_rate": 7.833851983944916e-06,
      "loss": 0.7332,
      "step": 2190
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.369900226593018,
      "learning_rate": 7.831847403501253e-06,
      "loss": 0.4772,
      "step": 2191
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.348221778869629,
      "learning_rate": 7.829842152697717e-06,
      "loss": 0.5526,
      "step": 2192
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5258984565734863,
      "learning_rate": 7.827836232008992e-06,
      "loss": 0.4441,
      "step": 2193
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.301513671875,
      "learning_rate": 7.825829641909923e-06,
      "loss": 0.4716,
      "step": 2194
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4209535121917725,
      "learning_rate": 7.823822382875514e-06,
      "loss": 0.4124,
      "step": 2195
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4928820133209229,
      "learning_rate": 7.821814455380927e-06,
      "loss": 0.482,
      "step": 2196
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.481377124786377,
      "learning_rate": 7.819805859901483e-06,
      "loss": 0.5503,
      "step": 2197
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0526483058929443,
      "learning_rate": 7.817796596912658e-06,
      "loss": 0.4545,
      "step": 2198
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7270750999450684,
      "learning_rate": 7.81578666689009e-06,
      "loss": 0.4508,
      "step": 2199
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.637575626373291,
      "learning_rate": 7.813776070309573e-06,
      "loss": 0.4667,
      "step": 2200
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.310107707977295,
      "learning_rate": 7.811764807647059e-06,
      "loss": 0.4192,
      "step": 2201
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5780596733093262,
      "learning_rate": 7.809752879378655e-06,
      "loss": 0.4457,
      "step": 2202
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.422376275062561,
      "learning_rate": 7.807740285980633e-06,
      "loss": 0.489,
      "step": 2203
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7778304815292358,
      "learning_rate": 7.805727027929413e-06,
      "loss": 0.5543,
      "step": 2204
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2050193548202515,
      "learning_rate": 7.80371310570158e-06,
      "loss": 0.4507,
      "step": 2205
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6630326509475708,
      "learning_rate": 7.801698519773872e-06,
      "loss": 0.5151,
      "step": 2206
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8248732089996338,
      "learning_rate": 7.799683270623186e-06,
      "loss": 0.6049,
      "step": 2207
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4445621967315674,
      "learning_rate": 7.797667358726576e-06,
      "loss": 0.463,
      "step": 2208
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.2174673080444336,
      "learning_rate": 7.795650784561249e-06,
      "loss": 0.4003,
      "step": 2209
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6684212684631348,
      "learning_rate": 7.793633548604573e-06,
      "loss": 0.5051,
      "step": 2210
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5861806869506836,
      "learning_rate": 7.791615651334075e-06,
      "loss": 0.5262,
      "step": 2211
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8645062446594238,
      "learning_rate": 7.789597093227432e-06,
      "loss": 0.5076,
      "step": 2212
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4245061874389648,
      "learning_rate": 7.78757787476248e-06,
      "loss": 0.5564,
      "step": 2213
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5520128011703491,
      "learning_rate": 7.785557996417214e-06,
      "loss": 0.5061,
      "step": 2214
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.618715286254883,
      "learning_rate": 7.783537458669783e-06,
      "loss": 0.469,
      "step": 2215
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4093191623687744,
      "learning_rate": 7.781516261998492e-06,
      "loss": 0.5406,
      "step": 2216
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0788941383361816,
      "learning_rate": 7.779494406881801e-06,
      "loss": 0.5222,
      "step": 2217
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4179260730743408,
      "learning_rate": 7.777471893798328e-06,
      "loss": 0.549,
      "step": 2218
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3942945003509521,
      "learning_rate": 7.775448723226846e-06,
      "loss": 0.4527,
      "step": 2219
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2102153301239014,
      "learning_rate": 7.773424895646284e-06,
      "loss": 0.72,
      "step": 2220
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6025006771087646,
      "learning_rate": 7.771400411535726e-06,
      "loss": 0.4565,
      "step": 2221
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3532992601394653,
      "learning_rate": 7.769375271374408e-06,
      "loss": 0.4727,
      "step": 2222
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5505386590957642,
      "learning_rate": 7.76734947564173e-06,
      "loss": 0.5516,
      "step": 2223
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3084747791290283,
      "learning_rate": 7.765323024817237e-06,
      "loss": 0.4489,
      "step": 2224
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.670043468475342,
      "learning_rate": 7.763295919380637e-06,
      "loss": 0.4999,
      "step": 2225
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.335007905960083,
      "learning_rate": 7.761268159811792e-06,
      "loss": 0.4809,
      "step": 2226
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.805523157119751,
      "learning_rate": 7.759239746590712e-06,
      "loss": 0.5264,
      "step": 2227
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4260573387145996,
      "learning_rate": 7.757210680197569e-06,
      "loss": 0.4172,
      "step": 2228
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.150881290435791,
      "learning_rate": 7.755180961112686e-06,
      "loss": 0.4208,
      "step": 2229
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.483120322227478,
      "learning_rate": 7.753150589816543e-06,
      "loss": 0.4552,
      "step": 2230
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8625496625900269,
      "learning_rate": 7.751119566789772e-06,
      "loss": 0.7681,
      "step": 2231
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.585390567779541,
      "learning_rate": 7.749087892513161e-06,
      "loss": 0.4866,
      "step": 2232
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2704514265060425,
      "learning_rate": 7.747055567467652e-06,
      "loss": 0.4948,
      "step": 2233
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5007624626159668,
      "learning_rate": 7.74502259213434e-06,
      "loss": 0.5489,
      "step": 2234
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.814143419265747,
      "learning_rate": 7.742988966994474e-06,
      "loss": 0.4771,
      "step": 2235
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.506827473640442,
      "learning_rate": 7.740954692529457e-06,
      "loss": 0.4569,
      "step": 2236
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5900578498840332,
      "learning_rate": 7.738919769220847e-06,
      "loss": 0.5302,
      "step": 2237
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6477731466293335,
      "learning_rate": 7.736884197550353e-06,
      "loss": 0.4651,
      "step": 2238
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2054520845413208,
      "learning_rate": 7.734847977999843e-06,
      "loss": 0.5819,
      "step": 2239
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.054335832595825,
      "learning_rate": 7.732811111051329e-06,
      "loss": 0.5246,
      "step": 2240
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4955378770828247,
      "learning_rate": 7.730773597186986e-06,
      "loss": 0.5441,
      "step": 2241
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.48868727684021,
      "learning_rate": 7.728735436889137e-06,
      "loss": 0.5042,
      "step": 2242
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4111701250076294,
      "learning_rate": 7.726696630640255e-06,
      "loss": 0.5142,
      "step": 2243
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2985652685165405,
      "learning_rate": 7.724657178922976e-06,
      "loss": 0.7736,
      "step": 2244
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2568702697753906,
      "learning_rate": 7.722617082220081e-06,
      "loss": 0.7164,
      "step": 2245
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.560396432876587,
      "learning_rate": 7.7205763410145e-06,
      "loss": 0.5136,
      "step": 2246
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.826767086982727,
      "learning_rate": 7.718534955789327e-06,
      "loss": 0.5554,
      "step": 2247
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4937982559204102,
      "learning_rate": 7.716492927027799e-06,
      "loss": 0.4784,
      "step": 2248
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8799065351486206,
      "learning_rate": 7.714450255213311e-06,
      "loss": 0.5075,
      "step": 2249
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5639879703521729,
      "learning_rate": 7.712406940829404e-06,
      "loss": 0.5495,
      "step": 2250
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7592717409133911,
      "learning_rate": 7.710362984359781e-06,
      "loss": 0.5017,
      "step": 2251
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9350404739379883,
      "learning_rate": 7.708318386288283e-06,
      "loss": 0.4381,
      "step": 2252
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5477168560028076,
      "learning_rate": 7.706273147098917e-06,
      "loss": 0.4962,
      "step": 2253
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6947778463363647,
      "learning_rate": 7.704227267275834e-06,
      "loss": 0.4462,
      "step": 2254
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.1961750984191895,
      "learning_rate": 7.70218074730334e-06,
      "loss": 0.4091,
      "step": 2255
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.359955906867981,
      "learning_rate": 7.700133587665885e-06,
      "loss": 0.4341,
      "step": 2256
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.868581771850586,
      "learning_rate": 7.698085788848083e-06,
      "loss": 0.4505,
      "step": 2257
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2034674882888794,
      "learning_rate": 7.696037351334689e-06,
      "loss": 0.6974,
      "step": 2258
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3935621976852417,
      "learning_rate": 7.693988275610613e-06,
      "loss": 0.5045,
      "step": 2259
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3548355102539062,
      "learning_rate": 7.691938562160916e-06,
      "loss": 0.4905,
      "step": 2260
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2395213842391968,
      "learning_rate": 7.68988821147081e-06,
      "loss": 0.4896,
      "step": 2261
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1543610095977783,
      "learning_rate": 7.687837224025656e-06,
      "loss": 0.4405,
      "step": 2262
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1953182220458984,
      "learning_rate": 7.685785600310972e-06,
      "loss": 0.5578,
      "step": 2263
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4551104307174683,
      "learning_rate": 7.683733340812416e-06,
      "loss": 0.5606,
      "step": 2264
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3207252025604248,
      "learning_rate": 7.681680446015805e-06,
      "loss": 0.5128,
      "step": 2265
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2486639022827148,
      "learning_rate": 7.679626916407105e-06,
      "loss": 0.4384,
      "step": 2266
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3184739351272583,
      "learning_rate": 7.67757275247243e-06,
      "loss": 0.4861,
      "step": 2267
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2739629745483398,
      "learning_rate": 7.675517954698044e-06,
      "loss": 0.4485,
      "step": 2268
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3216806650161743,
      "learning_rate": 7.673462523570368e-06,
      "loss": 0.7776,
      "step": 2269
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2315773963928223,
      "learning_rate": 7.67140645957596e-06,
      "loss": 0.5136,
      "step": 2270
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2386504411697388,
      "learning_rate": 7.669349763201536e-06,
      "loss": 0.433,
      "step": 2271
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3179785013198853,
      "learning_rate": 7.667292434933966e-06,
      "loss": 0.5577,
      "step": 2272
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2475773096084595,
      "learning_rate": 7.665234475260262e-06,
      "loss": 0.4482,
      "step": 2273
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3228704929351807,
      "learning_rate": 7.663175884667585e-06,
      "loss": 0.4617,
      "step": 2274
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3694757223129272,
      "learning_rate": 7.66111666364325e-06,
      "loss": 0.4914,
      "step": 2275
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7341684103012085,
      "learning_rate": 7.659056812674722e-06,
      "loss": 0.428,
      "step": 2276
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.928049564361572,
      "learning_rate": 7.656996332249607e-06,
      "loss": 0.4606,
      "step": 2277
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1783089637756348,
      "learning_rate": 7.65493522285567e-06,
      "loss": 0.4808,
      "step": 2278
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1927298307418823,
      "learning_rate": 7.65287348498082e-06,
      "loss": 0.4215,
      "step": 2279
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5040966272354126,
      "learning_rate": 7.650811119113112e-06,
      "loss": 0.5291,
      "step": 2280
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4951666593551636,
      "learning_rate": 7.648748125740757e-06,
      "loss": 0.414,
      "step": 2281
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2738624811172485,
      "learning_rate": 7.646684505352108e-06,
      "loss": 0.4463,
      "step": 2282
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7360789775848389,
      "learning_rate": 7.64462025843567e-06,
      "loss": 0.4983,
      "step": 2283
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3499791622161865,
      "learning_rate": 7.642555385480095e-06,
      "loss": 0.496,
      "step": 2284
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3307958841323853,
      "learning_rate": 7.64048988697418e-06,
      "loss": 0.4329,
      "step": 2285
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4622972011566162,
      "learning_rate": 7.638423763406878e-06,
      "loss": 0.5449,
      "step": 2286
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1259405612945557,
      "learning_rate": 7.636357015267284e-06,
      "loss": 0.4105,
      "step": 2287
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4388898611068726,
      "learning_rate": 7.634289643044643e-06,
      "loss": 0.53,
      "step": 2288
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.514983892440796,
      "learning_rate": 7.632221647228345e-06,
      "loss": 0.5398,
      "step": 2289
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8252805471420288,
      "learning_rate": 7.63015302830793e-06,
      "loss": 0.5177,
      "step": 2290
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3985956907272339,
      "learning_rate": 7.628083786773084e-06,
      "loss": 0.481,
      "step": 2291
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7401392459869385,
      "learning_rate": 7.626013923113644e-06,
      "loss": 0.5293,
      "step": 2292
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4538785219192505,
      "learning_rate": 7.623943437819592e-06,
      "loss": 0.3825,
      "step": 2293
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4249389171600342,
      "learning_rate": 7.621872331381058e-06,
      "loss": 0.4204,
      "step": 2294
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4469090700149536,
      "learning_rate": 7.619800604288312e-06,
      "loss": 0.5052,
      "step": 2295
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4354431629180908,
      "learning_rate": 7.61772825703178e-06,
      "loss": 0.4318,
      "step": 2296
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.087880849838257,
      "learning_rate": 7.615655290102033e-06,
      "loss": 0.4636,
      "step": 2297
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3987025022506714,
      "learning_rate": 7.613581703989787e-06,
      "loss": 0.4677,
      "step": 2298
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.149223804473877,
      "learning_rate": 7.611507499185905e-06,
      "loss": 0.4504,
      "step": 2299
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3444814682006836,
      "learning_rate": 7.609432676181396e-06,
      "loss": 0.4678,
      "step": 2300
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3409909009933472,
      "learning_rate": 7.607357235467417e-06,
      "loss": 0.465,
      "step": 2301
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0806854963302612,
      "learning_rate": 7.605281177535264e-06,
      "loss": 0.429,
      "step": 2302
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2339870929718018,
      "learning_rate": 7.603204502876393e-06,
      "loss": 0.4583,
      "step": 2303
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3485418558120728,
      "learning_rate": 7.601127211982393e-06,
      "loss": 0.7418,
      "step": 2304
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5066558122634888,
      "learning_rate": 7.5990493053450076e-06,
      "loss": 0.5079,
      "step": 2305
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.5012991428375244,
      "learning_rate": 7.596970783456119e-06,
      "loss": 0.4726,
      "step": 2306
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.412563681602478,
      "learning_rate": 7.5948916468077606e-06,
      "loss": 0.5223,
      "step": 2307
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2095954418182373,
      "learning_rate": 7.592811895892108e-06,
      "loss": 0.5208,
      "step": 2308
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8435156345367432,
      "learning_rate": 7.590731531201485e-06,
      "loss": 0.441,
      "step": 2309
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9638692140579224,
      "learning_rate": 7.588650553228358e-06,
      "loss": 0.4807,
      "step": 2310
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5371395349502563,
      "learning_rate": 7.586568962465341e-06,
      "loss": 0.4799,
      "step": 2311
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5414412021636963,
      "learning_rate": 7.584486759405191e-06,
      "loss": 0.5195,
      "step": 2312
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.471550703048706,
      "learning_rate": 7.582403944540808e-06,
      "loss": 0.4592,
      "step": 2313
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.0746207237243652,
      "learning_rate": 7.580320518365245e-06,
      "loss": 0.4989,
      "step": 2314
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6572483777999878,
      "learning_rate": 7.57823648137169e-06,
      "loss": 0.4951,
      "step": 2315
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4454314708709717,
      "learning_rate": 7.5761518340534825e-06,
      "loss": 0.5325,
      "step": 2316
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.303371787071228,
      "learning_rate": 7.5740665769041026e-06,
      "loss": 0.4695,
      "step": 2317
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6290451288223267,
      "learning_rate": 7.5719807104171775e-06,
      "loss": 0.57,
      "step": 2318
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2372182607650757,
      "learning_rate": 7.569894235086472e-06,
      "loss": 0.7624,
      "step": 2319
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.160767912864685,
      "learning_rate": 7.567807151405908e-06,
      "loss": 0.5311,
      "step": 2320
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.297935128211975,
      "learning_rate": 7.565719459869538e-06,
      "loss": 0.389,
      "step": 2321
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.139345645904541,
      "learning_rate": 7.5636311609715665e-06,
      "loss": 0.4064,
      "step": 2322
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.348400592803955,
      "learning_rate": 7.561542255206338e-06,
      "loss": 0.4653,
      "step": 2323
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.810332179069519,
      "learning_rate": 7.559452743068342e-06,
      "loss": 0.4289,
      "step": 2324
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5292776823043823,
      "learning_rate": 7.557362625052211e-06,
      "loss": 0.4881,
      "step": 2325
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.370377540588379,
      "learning_rate": 7.555271901652724e-06,
      "loss": 0.4948,
      "step": 2326
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9211256504058838,
      "learning_rate": 7.553180573364799e-06,
      "loss": 0.4311,
      "step": 2327
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3292717933654785,
      "learning_rate": 7.551088640683498e-06,
      "loss": 0.4988,
      "step": 2328
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.442737102508545,
      "learning_rate": 7.548996104104029e-06,
      "loss": 0.4584,
      "step": 2329
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3522764444351196,
      "learning_rate": 7.546902964121739e-06,
      "loss": 0.4503,
      "step": 2330
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4097334146499634,
      "learning_rate": 7.544809221232119e-06,
      "loss": 0.5189,
      "step": 2331
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4976346492767334,
      "learning_rate": 7.542714875930807e-06,
      "loss": 0.5162,
      "step": 2332
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5527530908584595,
      "learning_rate": 7.5406199287135775e-06,
      "loss": 0.495,
      "step": 2333
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8676705360412598,
      "learning_rate": 7.5385243800763505e-06,
      "loss": 0.5453,
      "step": 2334
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4243558645248413,
      "learning_rate": 7.536428230515188e-06,
      "loss": 0.4368,
      "step": 2335
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.505013942718506,
      "learning_rate": 7.5343314805262945e-06,
      "loss": 0.4298,
      "step": 2336
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.39082932472229,
      "learning_rate": 7.532234130606015e-06,
      "loss": 0.5138,
      "step": 2337
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.496479868888855,
      "learning_rate": 7.53013618125084e-06,
      "loss": 0.4734,
      "step": 2338
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2996288537979126,
      "learning_rate": 7.528037632957399e-06,
      "loss": 0.4353,
      "step": 2339
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5359388589859009,
      "learning_rate": 7.525938486222464e-06,
      "loss": 0.51,
      "step": 2340
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.528822898864746,
      "learning_rate": 7.523838741542948e-06,
      "loss": 0.4646,
      "step": 2341
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4587113857269287,
      "learning_rate": 7.5217383994159054e-06,
      "loss": 0.4535,
      "step": 2342
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4063305854797363,
      "learning_rate": 7.519637460338536e-06,
      "loss": 0.4381,
      "step": 2343
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3379665613174438,
      "learning_rate": 7.517535924808174e-06,
      "loss": 0.4864,
      "step": 2344
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.41837739944458,
      "learning_rate": 7.515433793322302e-06,
      "loss": 0.5117,
      "step": 2345
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8851724863052368,
      "learning_rate": 7.513331066378538e-06,
      "loss": 0.453,
      "step": 2346
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2096350193023682,
      "learning_rate": 7.511227744474645e-06,
      "loss": 0.7256,
      "step": 2347
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3030955791473389,
      "learning_rate": 7.509123828108523e-06,
      "loss": 0.5016,
      "step": 2348
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1342936754226685,
      "learning_rate": 7.507019317778216e-06,
      "loss": 0.5242,
      "step": 2349
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5354493856430054,
      "learning_rate": 7.5049142139819066e-06,
      "loss": 0.4846,
      "step": 2350
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.475203037261963,
      "learning_rate": 7.50280851721792e-06,
      "loss": 0.5177,
      "step": 2351
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5536659955978394,
      "learning_rate": 7.500702227984718e-06,
      "loss": 0.4489,
      "step": 2352
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5496673583984375,
      "learning_rate": 7.498595346780908e-06,
      "loss": 0.5641,
      "step": 2353
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2328875064849854,
      "learning_rate": 7.496487874105234e-06,
      "loss": 0.7429,
      "step": 2354
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2759519815444946,
      "learning_rate": 7.494379810456578e-06,
      "loss": 0.4569,
      "step": 2355
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6192456483840942,
      "learning_rate": 7.4922711563339675e-06,
      "loss": 0.4926,
      "step": 2356
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5140470266342163,
      "learning_rate": 7.490161912236564e-06,
      "loss": 0.449,
      "step": 2357
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8537524938583374,
      "learning_rate": 7.488052078663674e-06,
      "loss": 0.4417,
      "step": 2358
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7635010480880737,
      "learning_rate": 7.485941656114742e-06,
      "loss": 0.4341,
      "step": 2359
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.053107500076294,
      "learning_rate": 7.483830645089348e-06,
      "loss": 0.4703,
      "step": 2360
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5880178213119507,
      "learning_rate": 7.481719046087216e-06,
      "loss": 0.4425,
      "step": 2361
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2854012250900269,
      "learning_rate": 7.479606859608205e-06,
      "loss": 0.4636,
      "step": 2362
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.0576202869415283,
      "learning_rate": 7.477494086152317e-06,
      "loss": 0.4771,
      "step": 2363
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2836523056030273,
      "learning_rate": 7.475380726219692e-06,
      "loss": 0.4865,
      "step": 2364
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4561231136322021,
      "learning_rate": 7.473266780310608e-06,
      "loss": 0.5223,
      "step": 2365
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3260040283203125,
      "learning_rate": 7.471152248925484e-06,
      "loss": 0.4563,
      "step": 2366
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.358450174331665,
      "learning_rate": 7.4690371325648714e-06,
      "loss": 0.4792,
      "step": 2367
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6175297498703003,
      "learning_rate": 7.466921431729467e-06,
      "loss": 0.5105,
      "step": 2368
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.499531626701355,
      "learning_rate": 7.464805146920101e-06,
      "loss": 0.5301,
      "step": 2369
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2346210479736328,
      "learning_rate": 7.462688278637746e-06,
      "loss": 0.698,
      "step": 2370
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7540225982666016,
      "learning_rate": 7.460570827383511e-06,
      "loss": 0.4896,
      "step": 2371
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2038283348083496,
      "learning_rate": 7.4584527936586435e-06,
      "loss": 0.4605,
      "step": 2372
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2354705333709717,
      "learning_rate": 7.456334177964525e-06,
      "loss": 0.4774,
      "step": 2373
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4046008586883545,
      "learning_rate": 7.454214980802678e-06,
      "loss": 0.5101,
      "step": 2374
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0715947151184082,
      "learning_rate": 7.452095202674766e-06,
      "loss": 0.7047,
      "step": 2375
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4044249057769775,
      "learning_rate": 7.4499748440825845e-06,
      "loss": 0.4587,
      "step": 2376
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6338887214660645,
      "learning_rate": 7.447853905528069e-06,
      "loss": 0.4871,
      "step": 2377
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4512293338775635,
      "learning_rate": 7.445732387513292e-06,
      "loss": 0.464,
      "step": 2378
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1808197498321533,
      "learning_rate": 7.443610290540463e-06,
      "loss": 0.4537,
      "step": 2379
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0586307048797607,
      "learning_rate": 7.441487615111926e-06,
      "loss": 0.4556,
      "step": 2380
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0703823566436768,
      "learning_rate": 7.439364361730167e-06,
      "loss": 0.493,
      "step": 2381
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3521071672439575,
      "learning_rate": 7.437240530897807e-06,
      "loss": 0.4796,
      "step": 2382
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5243695974349976,
      "learning_rate": 7.435116123117601e-06,
      "loss": 0.4135,
      "step": 2383
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1202268600463867,
      "learning_rate": 7.432991138892444e-06,
      "loss": 0.4429,
      "step": 2384
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6625473499298096,
      "learning_rate": 7.430865578725362e-06,
      "loss": 0.4906,
      "step": 2385
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9688929319381714,
      "learning_rate": 7.428739443119528e-06,
      "loss": 0.5588,
      "step": 2386
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.264723300933838,
      "learning_rate": 7.42661273257824e-06,
      "loss": 0.4366,
      "step": 2387
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6860084533691406,
      "learning_rate": 7.424485447604937e-06,
      "loss": 0.4797,
      "step": 2388
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4869558811187744,
      "learning_rate": 7.422357588703195e-06,
      "loss": 0.5244,
      "step": 2389
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.709638237953186,
      "learning_rate": 7.420229156376725e-06,
      "loss": 0.5062,
      "step": 2390
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.152063012123108,
      "learning_rate": 7.4181001511293694e-06,
      "loss": 0.7549,
      "step": 2391
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.9316771030426025,
      "learning_rate": 7.415970573465115e-06,
      "loss": 0.4773,
      "step": 2392
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.398383617401123,
      "learning_rate": 7.413840423888075e-06,
      "loss": 0.4927,
      "step": 2393
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3614553213119507,
      "learning_rate": 7.4117097029025055e-06,
      "loss": 0.5037,
      "step": 2394
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.355072021484375,
      "learning_rate": 7.409578411012792e-06,
      "loss": 0.3809,
      "step": 2395
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3679708242416382,
      "learning_rate": 7.407446548723458e-06,
      "loss": 0.4929,
      "step": 2396
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7034530639648438,
      "learning_rate": 7.405314116539164e-06,
      "loss": 0.4879,
      "step": 2397
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8319326639175415,
      "learning_rate": 7.4031811149647015e-06,
      "loss": 0.443,
      "step": 2398
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8337382078170776,
      "learning_rate": 7.4010475445049964e-06,
      "loss": 0.4407,
      "step": 2399
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3948476314544678,
      "learning_rate": 7.398913405665115e-06,
      "loss": 0.4312,
      "step": 2400
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2206326723098755,
      "learning_rate": 7.396778698950251e-06,
      "loss": 0.4604,
      "step": 2401
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4471319913864136,
      "learning_rate": 7.394643424865737e-06,
      "loss": 0.4874,
      "step": 2402
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4629448652267456,
      "learning_rate": 7.392507583917041e-06,
      "loss": 0.4979,
      "step": 2403
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.044447898864746,
      "learning_rate": 7.3903711766097604e-06,
      "loss": 0.4453,
      "step": 2404
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8816602230072021,
      "learning_rate": 7.38823420344963e-06,
      "loss": 0.4715,
      "step": 2405
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2883158922195435,
      "learning_rate": 7.386096664942518e-06,
      "loss": 0.5359,
      "step": 2406
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2391704320907593,
      "learning_rate": 7.383958561594426e-06,
      "loss": 0.4618,
      "step": 2407
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2998210191726685,
      "learning_rate": 7.381819893911491e-06,
      "loss": 0.4655,
      "step": 2408
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5660020112991333,
      "learning_rate": 7.3796806623999785e-06,
      "loss": 0.518,
      "step": 2409
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2920161485671997,
      "learning_rate": 7.377540867566295e-06,
      "loss": 0.5116,
      "step": 2410
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4817880392074585,
      "learning_rate": 7.375400509916975e-06,
      "loss": 0.5095,
      "step": 2411
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.831212043762207,
      "learning_rate": 7.373259589958688e-06,
      "loss": 0.3916,
      "step": 2412
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8610485792160034,
      "learning_rate": 7.371118108198234e-06,
      "loss": 0.5314,
      "step": 2413
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2821162939071655,
      "learning_rate": 7.368976065142552e-06,
      "loss": 0.4662,
      "step": 2414
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6589787006378174,
      "learning_rate": 7.366833461298709e-06,
      "loss": 0.452,
      "step": 2415
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.453816294670105,
      "learning_rate": 7.364690297173904e-06,
      "loss": 0.4732,
      "step": 2416
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5245256423950195,
      "learning_rate": 7.362546573275473e-06,
      "loss": 0.4704,
      "step": 2417
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2122868299484253,
      "learning_rate": 7.360402290110882e-06,
      "loss": 0.505,
      "step": 2418
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4016951322555542,
      "learning_rate": 7.358257448187726e-06,
      "loss": 0.5243,
      "step": 2419
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.585078239440918,
      "learning_rate": 7.356112048013741e-06,
      "loss": 0.5134,
      "step": 2420
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4736675024032593,
      "learning_rate": 7.353966090096786e-06,
      "loss": 0.4975,
      "step": 2421
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.726965308189392,
      "learning_rate": 7.3518195749448565e-06,
      "loss": 0.5108,
      "step": 2422
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.725001573562622,
      "learning_rate": 7.349672503066081e-06,
      "loss": 0.4371,
      "step": 2423
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3574092388153076,
      "learning_rate": 7.347524874968716e-06,
      "loss": 0.5275,
      "step": 2424
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0069074630737305,
      "learning_rate": 7.3453766911611525e-06,
      "loss": 0.4362,
      "step": 2425
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7783420085906982,
      "learning_rate": 7.3432279521519145e-06,
      "loss": 0.5239,
      "step": 2426
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.957133412361145,
      "learning_rate": 7.341078658449653e-06,
      "loss": 0.5339,
      "step": 2427
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9964228868484497,
      "learning_rate": 7.338928810563152e-06,
      "loss": 0.4856,
      "step": 2428
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.658233642578125,
      "learning_rate": 7.336778409001329e-06,
      "loss": 0.5126,
      "step": 2429
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6218528747558594,
      "learning_rate": 7.334627454273229e-06,
      "loss": 0.5225,
      "step": 2430
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.429949164390564,
      "learning_rate": 7.332475946888031e-06,
      "loss": 0.4774,
      "step": 2431
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.310510516166687,
      "learning_rate": 7.330323887355044e-06,
      "loss": 0.4952,
      "step": 2432
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4262596368789673,
      "learning_rate": 7.328171276183708e-06,
      "loss": 0.4578,
      "step": 2433
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5119045972824097,
      "learning_rate": 7.32601811388359e-06,
      "loss": 0.5055,
      "step": 2434
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4300642013549805,
      "learning_rate": 7.323864400964391e-06,
      "loss": 0.3792,
      "step": 2435
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8812023401260376,
      "learning_rate": 7.321710137935944e-06,
      "loss": 0.4748,
      "step": 2436
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4702354669570923,
      "learning_rate": 7.319555325308207e-06,
      "loss": 0.4592,
      "step": 2437
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9636346101760864,
      "learning_rate": 7.317399963591276e-06,
      "loss": 0.4892,
      "step": 2438
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6122008562088013,
      "learning_rate": 7.315244053295366e-06,
      "loss": 0.5111,
      "step": 2439
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6103947162628174,
      "learning_rate": 7.313087594930831e-06,
      "loss": 0.4473,
      "step": 2440
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8980401754379272,
      "learning_rate": 7.310930589008151e-06,
      "loss": 0.5506,
      "step": 2441
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.171992301940918,
      "learning_rate": 7.308773036037935e-06,
      "loss": 0.5284,
      "step": 2442
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5367882251739502,
      "learning_rate": 7.306614936530926e-06,
      "loss": 0.5619,
      "step": 2443
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.613923192024231,
      "learning_rate": 7.304456290997991e-06,
      "loss": 0.4944,
      "step": 2444
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8557815551757812,
      "learning_rate": 7.302297099950127e-06,
      "loss": 0.4922,
      "step": 2445
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.1110050678253174,
      "learning_rate": 7.300137363898464e-06,
      "loss": 0.5211,
      "step": 2446
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4665138721466064,
      "learning_rate": 7.2979770833542565e-06,
      "loss": 0.5102,
      "step": 2447
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5921279191970825,
      "learning_rate": 7.295816258828892e-06,
      "loss": 0.4918,
      "step": 2448
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5941818952560425,
      "learning_rate": 7.2936548908338825e-06,
      "loss": 0.5251,
      "step": 2449
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6195759773254395,
      "learning_rate": 7.291492979880872e-06,
      "loss": 0.4529,
      "step": 2450
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.034752607345581,
      "learning_rate": 7.289330526481633e-06,
      "loss": 0.4617,
      "step": 2451
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3517276048660278,
      "learning_rate": 7.287167531148061e-06,
      "loss": 0.4855,
      "step": 2452
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8415199518203735,
      "learning_rate": 7.285003994392188e-06,
      "loss": 0.5524,
      "step": 2453
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3622664213180542,
      "learning_rate": 7.28283991672617e-06,
      "loss": 0.4712,
      "step": 2454
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.419535756111145,
      "learning_rate": 7.2806752986622895e-06,
      "loss": 0.4627,
      "step": 2455
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8865125179290771,
      "learning_rate": 7.278510140712961e-06,
      "loss": 0.4957,
      "step": 2456
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.2071502208709717,
      "learning_rate": 7.2763444433907216e-06,
      "loss": 0.5046,
      "step": 2457
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4195029735565186,
      "learning_rate": 7.2741782072082405e-06,
      "loss": 0.5129,
      "step": 2458
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3767067193984985,
      "learning_rate": 7.272011432678313e-06,
      "loss": 0.4578,
      "step": 2459
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5734814405441284,
      "learning_rate": 7.2698441203138615e-06,
      "loss": 0.4708,
      "step": 2460
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3659517765045166,
      "learning_rate": 7.267676270627937e-06,
      "loss": 0.5104,
      "step": 2461
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4920942783355713,
      "learning_rate": 7.2655078841337155e-06,
      "loss": 0.467,
      "step": 2462
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.1609997749328613,
      "learning_rate": 7.263338961344499e-06,
      "loss": 0.5586,
      "step": 2463
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6769886016845703,
      "learning_rate": 7.261169502773724e-06,
      "loss": 0.5079,
      "step": 2464
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.1367623805999756,
      "learning_rate": 7.2589995089349454e-06,
      "loss": 0.5241,
      "step": 2465
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4908335208892822,
      "learning_rate": 7.256828980341847e-06,
      "loss": 0.5126,
      "step": 2466
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3641585111618042,
      "learning_rate": 7.2546579175082425e-06,
      "loss": 0.7303,
      "step": 2467
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.1140530109405518,
      "learning_rate": 7.252486320948068e-06,
      "loss": 0.5103,
      "step": 2468
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.2885217666625977,
      "learning_rate": 7.2503141911753885e-06,
      "loss": 0.5197,
      "step": 2469
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.297018051147461,
      "learning_rate": 7.2481415287043934e-06,
      "loss": 0.4398,
      "step": 2470
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.30542254447937,
      "learning_rate": 7.2459683340494e-06,
      "loss": 0.4276,
      "step": 2471
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.551504135131836,
      "learning_rate": 7.2437946077248514e-06,
      "loss": 0.5051,
      "step": 2472
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5641589164733887,
      "learning_rate": 7.241620350245314e-06,
      "loss": 0.4868,
      "step": 2473
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.690780520439148,
      "learning_rate": 7.239445562125484e-06,
      "loss": 0.5216,
      "step": 2474
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.631860613822937,
      "learning_rate": 7.237270243880179e-06,
      "loss": 0.4913,
      "step": 2475
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3997267484664917,
      "learning_rate": 7.235094396024345e-06,
      "loss": 0.4331,
      "step": 2476
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.476111650466919,
      "learning_rate": 7.232918019073054e-06,
      "loss": 0.4577,
      "step": 2477
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5413538217544556,
      "learning_rate": 7.230741113541499e-06,
      "loss": 0.5027,
      "step": 2478
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3349597454071045,
      "learning_rate": 7.228563679945004e-06,
      "loss": 0.4974,
      "step": 2479
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8757579326629639,
      "learning_rate": 7.226385718799011e-06,
      "loss": 0.4616,
      "step": 2480
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1786025762557983,
      "learning_rate": 7.224207230619095e-06,
      "loss": 0.7528,
      "step": 2481
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4051212072372437,
      "learning_rate": 7.2220282159209485e-06,
      "loss": 0.5349,
      "step": 2482
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3618839979171753,
      "learning_rate": 7.2198486752203924e-06,
      "loss": 0.4115,
      "step": 2483
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5100210905075073,
      "learning_rate": 7.217668609033372e-06,
      "loss": 0.5114,
      "step": 2484
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.2651548385620117,
      "learning_rate": 7.2154880178759556e-06,
      "loss": 0.4925,
      "step": 2485
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5878945589065552,
      "learning_rate": 7.213306902264338e-06,
      "loss": 0.5151,
      "step": 2486
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4251011610031128,
      "learning_rate": 7.211125262714835e-06,
      "loss": 0.5159,
      "step": 2487
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.630764961242676,
      "learning_rate": 7.208943099743888e-06,
      "loss": 0.4467,
      "step": 2488
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1375749111175537,
      "learning_rate": 7.2067604138680636e-06,
      "loss": 0.4016,
      "step": 2489
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7486096620559692,
      "learning_rate": 7.204577205604051e-06,
      "loss": 0.4192,
      "step": 2490
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7486231327056885,
      "learning_rate": 7.202393475468663e-06,
      "loss": 0.5403,
      "step": 2491
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.111201763153076,
      "learning_rate": 7.200209223978836e-06,
      "loss": 0.4828,
      "step": 2492
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2950770854949951,
      "learning_rate": 7.1980244516516295e-06,
      "loss": 0.7766,
      "step": 2493
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6323091983795166,
      "learning_rate": 7.195839159004227e-06,
      "loss": 0.49,
      "step": 2494
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6245359182357788,
      "learning_rate": 7.1936533465539335e-06,
      "loss": 0.461,
      "step": 2495
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5538345575332642,
      "learning_rate": 7.19146701481818e-06,
      "loss": 0.4921,
      "step": 2496
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6439069509506226,
      "learning_rate": 7.189280164314517e-06,
      "loss": 0.5371,
      "step": 2497
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.650901198387146,
      "learning_rate": 7.187092795560623e-06,
      "loss": 0.483,
      "step": 2498
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.147801160812378,
      "learning_rate": 7.184904909074293e-06,
      "loss": 0.4686,
      "step": 2499
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8326590061187744,
      "learning_rate": 7.182716505373447e-06,
      "loss": 0.5501,
      "step": 2500
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5514826774597168,
      "learning_rate": 7.1805275849761295e-06,
      "loss": 0.4426,
      "step": 2501
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7910784482955933,
      "learning_rate": 7.178338148400502e-06,
      "loss": 0.4271,
      "step": 2502
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4198063611984253,
      "learning_rate": 7.176148196164856e-06,
      "loss": 0.4888,
      "step": 2503
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2237935066223145,
      "learning_rate": 7.173957728787602e-06,
      "loss": 0.4616,
      "step": 2504
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.3044352531433105,
      "learning_rate": 7.171766746787265e-06,
      "loss": 0.4942,
      "step": 2505
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6798173189163208,
      "learning_rate": 7.169575250682503e-06,
      "loss": 0.4982,
      "step": 2506
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.366178035736084,
      "learning_rate": 7.167383240992089e-06,
      "loss": 0.4868,
      "step": 2507
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.543215036392212,
      "learning_rate": 7.16519071823492e-06,
      "loss": 0.5078,
      "step": 2508
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3262287378311157,
      "learning_rate": 7.162997682930013e-06,
      "loss": 0.406,
      "step": 2509
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3950371742248535,
      "learning_rate": 7.16080413559651e-06,
      "loss": 0.5104,
      "step": 2510
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2586897611618042,
      "learning_rate": 7.158610076753667e-06,
      "loss": 0.4616,
      "step": 2511
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3469206094741821,
      "learning_rate": 7.156415506920869e-06,
      "loss": 0.4583,
      "step": 2512
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6570427417755127,
      "learning_rate": 7.154220426617615e-06,
      "loss": 0.4946,
      "step": 2513
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2677804231643677,
      "learning_rate": 7.1520248363635315e-06,
      "loss": 0.4323,
      "step": 2514
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.0495545864105225,
      "learning_rate": 7.149828736678361e-06,
      "loss": 0.5392,
      "step": 2515
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1558339595794678,
      "learning_rate": 7.1476321280819695e-06,
      "loss": 0.7315,
      "step": 2516
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3308461904525757,
      "learning_rate": 7.1454350110943394e-06,
      "loss": 0.4512,
      "step": 2517
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4842052459716797,
      "learning_rate": 7.143237386235577e-06,
      "loss": 0.5536,
      "step": 2518
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.2793757915496826,
      "learning_rate": 7.141039254025908e-06,
      "loss": 0.5395,
      "step": 2519
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.10318922996521,
      "learning_rate": 7.138840614985681e-06,
      "loss": 0.5154,
      "step": 2520
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.245764970779419,
      "learning_rate": 7.136641469635357e-06,
      "loss": 0.5499,
      "step": 2521
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7759509086608887,
      "learning_rate": 7.134441818495526e-06,
      "loss": 0.5323,
      "step": 2522
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7051454782485962,
      "learning_rate": 7.132241662086889e-06,
      "loss": 0.5276,
      "step": 2523
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3286758661270142,
      "learning_rate": 7.1300410009302726e-06,
      "loss": 0.7521,
      "step": 2524
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7543044090270996,
      "learning_rate": 7.127839835546623e-06,
      "loss": 0.5056,
      "step": 2525
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2369678020477295,
      "learning_rate": 7.125638166457e-06,
      "loss": 0.432,
      "step": 2526
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2280941009521484,
      "learning_rate": 7.123435994182592e-06,
      "loss": 0.5035,
      "step": 2527
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.614170551300049,
      "learning_rate": 7.121233319244697e-06,
      "loss": 0.5132,
      "step": 2528
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6382242441177368,
      "learning_rate": 7.119030142164737e-06,
      "loss": 0.5071,
      "step": 2529
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3545761108398438,
      "learning_rate": 7.1168264634642504e-06,
      "loss": 0.4757,
      "step": 2530
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4492374658584595,
      "learning_rate": 7.114622283664898e-06,
      "loss": 0.4065,
      "step": 2531
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4340574741363525,
      "learning_rate": 7.112417603288457e-06,
      "loss": 0.4212,
      "step": 2532
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4369606971740723,
      "learning_rate": 7.1102124228568235e-06,
      "loss": 0.5286,
      "step": 2533
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3895667791366577,
      "learning_rate": 7.108006742892012e-06,
      "loss": 0.6177,
      "step": 2534
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.542954683303833,
      "learning_rate": 7.105800563916152e-06,
      "loss": 0.5132,
      "step": 2535
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.570797324180603,
      "learning_rate": 7.103593886451495e-06,
      "loss": 0.4615,
      "step": 2536
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2812682390213013,
      "learning_rate": 7.101386711020412e-06,
      "loss": 0.5396,
      "step": 2537
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9325299263000488,
      "learning_rate": 7.099179038145387e-06,
      "loss": 0.4859,
      "step": 2538
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5745413303375244,
      "learning_rate": 7.096970868349026e-06,
      "loss": 0.5314,
      "step": 2539
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4975804090499878,
      "learning_rate": 7.094762202154048e-06,
      "loss": 0.4663,
      "step": 2540
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5283570289611816,
      "learning_rate": 7.092553040083297e-06,
      "loss": 0.5273,
      "step": 2541
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3479560613632202,
      "learning_rate": 7.090343382659726e-06,
      "loss": 0.4624,
      "step": 2542
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3641115427017212,
      "learning_rate": 7.0881332304064085e-06,
      "loss": 0.4691,
      "step": 2543
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.708400845527649,
      "learning_rate": 7.085922583846539e-06,
      "loss": 0.5292,
      "step": 2544
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0772688388824463,
      "learning_rate": 7.083711443503423e-06,
      "loss": 0.5774,
      "step": 2545
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2086132764816284,
      "learning_rate": 7.081499809900485e-06,
      "loss": 0.4875,
      "step": 2546
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.425736904144287,
      "learning_rate": 7.079287683561269e-06,
      "loss": 0.5218,
      "step": 2547
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3314931392669678,
      "learning_rate": 7.0770750650094335e-06,
      "loss": 0.4866,
      "step": 2548
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5595695972442627,
      "learning_rate": 7.074861954768751e-06,
      "loss": 0.4895,
      "step": 2549
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5285687446594238,
      "learning_rate": 7.072648353363114e-06,
      "loss": 0.5134,
      "step": 2550
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1707319021224976,
      "learning_rate": 7.07043426131653e-06,
      "loss": 0.7391,
      "step": 2551
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2480043172836304,
      "learning_rate": 7.0682196791531235e-06,
      "loss": 0.4651,
      "step": 2552
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4984993934631348,
      "learning_rate": 7.066004607397133e-06,
      "loss": 0.5015,
      "step": 2553
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0873346328735352,
      "learning_rate": 7.0637890465729165e-06,
      "loss": 0.4581,
      "step": 2554
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2681856155395508,
      "learning_rate": 7.061572997204943e-06,
      "loss": 0.4389,
      "step": 2555
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.308622121810913,
      "learning_rate": 7.0593564598177996e-06,
      "loss": 0.4669,
      "step": 2556
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1227151155471802,
      "learning_rate": 7.057139434936191e-06,
      "loss": 0.4949,
      "step": 2557
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.092455506324768,
      "learning_rate": 7.054921923084933e-06,
      "loss": 0.4349,
      "step": 2558
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9799199104309082,
      "learning_rate": 7.052703924788961e-06,
      "loss": 0.4084,
      "step": 2559
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4705010652542114,
      "learning_rate": 7.050485440573321e-06,
      "loss": 0.5144,
      "step": 2560
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7397806644439697,
      "learning_rate": 7.048266470963179e-06,
      "loss": 0.4783,
      "step": 2561
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3523333072662354,
      "learning_rate": 7.046047016483811e-06,
      "loss": 0.5606,
      "step": 2562
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.770330548286438,
      "learning_rate": 7.043827077660612e-06,
      "loss": 0.5296,
      "step": 2563
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9449104070663452,
      "learning_rate": 7.0416066550190885e-06,
      "loss": 0.4988,
      "step": 2564
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8712596893310547,
      "learning_rate": 7.039385749084864e-06,
      "loss": 0.4949,
      "step": 2565
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3846367597579956,
      "learning_rate": 7.037164360383675e-06,
      "loss": 0.4938,
      "step": 2566
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8868032693862915,
      "learning_rate": 7.034942489441371e-06,
      "loss": 0.4542,
      "step": 2567
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.473646640777588,
      "learning_rate": 7.0327201367839195e-06,
      "loss": 0.4481,
      "step": 2568
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.516208291053772,
      "learning_rate": 7.030497302937396e-06,
      "loss": 0.5505,
      "step": 2569
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2515969276428223,
      "learning_rate": 7.028273988427999e-06,
      "loss": 0.4952,
      "step": 2570
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1291694641113281,
      "learning_rate": 7.026050193782032e-06,
      "loss": 0.4445,
      "step": 2571
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7802646160125732,
      "learning_rate": 7.023825919525916e-06,
      "loss": 0.4592,
      "step": 2572
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.349669337272644,
      "learning_rate": 7.021601166186185e-06,
      "loss": 0.4985,
      "step": 2573
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.153680443763733,
      "learning_rate": 7.019375934289487e-06,
      "loss": 0.4565,
      "step": 2574
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5394421815872192,
      "learning_rate": 7.017150224362581e-06,
      "loss": 0.5199,
      "step": 2575
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1651984453201294,
      "learning_rate": 7.014924036932346e-06,
      "loss": 0.7024,
      "step": 2576
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1754273176193237,
      "learning_rate": 7.012697372525763e-06,
      "loss": 0.4756,
      "step": 2577
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4857374429702759,
      "learning_rate": 7.010470231669935e-06,
      "loss": 0.4526,
      "step": 2578
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.610152244567871,
      "learning_rate": 7.008242614892074e-06,
      "loss": 0.4516,
      "step": 2579
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9900274276733398,
      "learning_rate": 7.006014522719503e-06,
      "loss": 0.4997,
      "step": 2580
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3463155031204224,
      "learning_rate": 7.003785955679665e-06,
      "loss": 0.4547,
      "step": 2581
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3672211170196533,
      "learning_rate": 7.001556914300108e-06,
      "loss": 0.4558,
      "step": 2582
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5303971767425537,
      "learning_rate": 6.999327399108493e-06,
      "loss": 0.4916,
      "step": 2583
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3448317050933838,
      "learning_rate": 6.997097410632596e-06,
      "loss": 0.5317,
      "step": 2584
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5121899843215942,
      "learning_rate": 6.994866949400304e-06,
      "loss": 0.4977,
      "step": 2585
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1623728275299072,
      "learning_rate": 6.992636015939614e-06,
      "loss": 0.7363,
      "step": 2586
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5985543727874756,
      "learning_rate": 6.99040461077864e-06,
      "loss": 0.4851,
      "step": 2587
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2106552124023438,
      "learning_rate": 6.988172734445601e-06,
      "loss": 0.7117,
      "step": 2588
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4878921508789062,
      "learning_rate": 6.9859403874688314e-06,
      "loss": 0.4723,
      "step": 2589
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.145200490951538,
      "learning_rate": 6.983707570376776e-06,
      "loss": 0.5028,
      "step": 2590
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6233973503112793,
      "learning_rate": 6.981474283697992e-06,
      "loss": 0.523,
      "step": 2591
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.476849913597107,
      "learning_rate": 6.979240527961147e-06,
      "loss": 0.4382,
      "step": 2592
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.772281527519226,
      "learning_rate": 6.97700630369502e-06,
      "loss": 0.482,
      "step": 2593
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3431930541992188,
      "learning_rate": 6.974771611428499e-06,
      "loss": 0.4797,
      "step": 2594
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5373609066009521,
      "learning_rate": 6.972536451690585e-06,
      "loss": 0.5045,
      "step": 2595
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4826524257659912,
      "learning_rate": 6.970300825010389e-06,
      "loss": 0.4246,
      "step": 2596
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.379397988319397,
      "learning_rate": 6.96806473191713e-06,
      "loss": 0.5492,
      "step": 2597
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5137251615524292,
      "learning_rate": 6.965828172940144e-06,
      "loss": 0.4225,
      "step": 2598
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.508824110031128,
      "learning_rate": 6.963591148608873e-06,
      "loss": 0.4264,
      "step": 2599
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2220351696014404,
      "learning_rate": 6.961353659452867e-06,
      "loss": 0.5282,
      "step": 2600
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.492920160293579,
      "learning_rate": 6.959115706001788e-06,
      "loss": 0.4932,
      "step": 2601
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.534513473510742,
      "learning_rate": 6.956877288785409e-06,
      "loss": 0.5419,
      "step": 2602
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.177581310272217,
      "learning_rate": 6.954638408333611e-06,
      "loss": 0.5473,
      "step": 2603
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3446072340011597,
      "learning_rate": 6.9523990651763896e-06,
      "loss": 0.4337,
      "step": 2604
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1383607387542725,
      "learning_rate": 6.950159259843843e-06,
      "loss": 0.511,
      "step": 2605
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2755075693130493,
      "learning_rate": 6.947918992866183e-06,
      "loss": 0.5235,
      "step": 2606
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5468471050262451,
      "learning_rate": 6.9456782647737275e-06,
      "loss": 0.458,
      "step": 2607
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3435616493225098,
      "learning_rate": 6.943437076096906e-06,
      "loss": 0.5246,
      "step": 2608
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1504340171813965,
      "learning_rate": 6.941195427366259e-06,
      "loss": 0.6884,
      "step": 2609
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0485968589782715,
      "learning_rate": 6.9389533191124315e-06,
      "loss": 0.5749,
      "step": 2610
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.783266544342041,
      "learning_rate": 6.93671075186618e-06,
      "loss": 0.4952,
      "step": 2611
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5237784385681152,
      "learning_rate": 6.934467726158368e-06,
      "loss": 0.4719,
      "step": 2612
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4162813425064087,
      "learning_rate": 6.9322242425199714e-06,
      "loss": 0.4419,
      "step": 2613
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6276240348815918,
      "learning_rate": 6.929980301482067e-06,
      "loss": 0.5071,
      "step": 2614
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4755587577819824,
      "learning_rate": 6.927735903575847e-06,
      "loss": 0.5374,
      "step": 2615
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3231905698776245,
      "learning_rate": 6.925491049332611e-06,
      "loss": 0.4897,
      "step": 2616
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5665501356124878,
      "learning_rate": 6.923245739283761e-06,
      "loss": 0.4075,
      "step": 2617
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3962914943695068,
      "learning_rate": 6.920999973960813e-06,
      "loss": 0.4704,
      "step": 2618
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.567458987236023,
      "learning_rate": 6.918753753895388e-06,
      "loss": 0.474,
      "step": 2619
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2043166160583496,
      "learning_rate": 6.916507079619217e-06,
      "loss": 0.4235,
      "step": 2620
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3048344850540161,
      "learning_rate": 6.914259951664133e-06,
      "loss": 0.4444,
      "step": 2621
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1651077270507812,
      "learning_rate": 6.912012370562082e-06,
      "loss": 0.3892,
      "step": 2622
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.449117422103882,
      "learning_rate": 6.909764336845116e-06,
      "loss": 0.5289,
      "step": 2623
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1133443117141724,
      "learning_rate": 6.907515851045392e-06,
      "loss": 0.7616,
      "step": 2624
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.8142282962799072,
      "learning_rate": 6.905266913695178e-06,
      "loss": 0.5439,
      "step": 2625
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3391666412353516,
      "learning_rate": 6.903017525326843e-06,
      "loss": 0.4569,
      "step": 2626
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1422899961471558,
      "learning_rate": 6.900767686472869e-06,
      "loss": 0.4126,
      "step": 2627
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4019745588302612,
      "learning_rate": 6.8985173976658395e-06,
      "loss": 0.4472,
      "step": 2628
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.200456976890564,
      "learning_rate": 6.896266659438449e-06,
      "loss": 0.7758,
      "step": 2629
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5852530002593994,
      "learning_rate": 6.894015472323494e-06,
      "loss": 0.432,
      "step": 2630
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4420697689056396,
      "learning_rate": 6.891763836853881e-06,
      "loss": 0.7598,
      "step": 2631
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3517509698867798,
      "learning_rate": 6.889511753562621e-06,
      "loss": 0.461,
      "step": 2632
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1889879703521729,
      "learning_rate": 6.887259222982829e-06,
      "loss": 0.4612,
      "step": 2633
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5292956829071045,
      "learning_rate": 6.885006245647732e-06,
      "loss": 0.4684,
      "step": 2634
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0928661823272705,
      "learning_rate": 6.882752822090656e-06,
      "loss": 0.7308,
      "step": 2635
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2192834615707397,
      "learning_rate": 6.880498952845035e-06,
      "loss": 0.4755,
      "step": 2636
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3666399717330933,
      "learning_rate": 6.87824463844441e-06,
      "loss": 0.4487,
      "step": 2637
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.646061658859253,
      "learning_rate": 6.8759898794224265e-06,
      "loss": 0.4746,
      "step": 2638
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7118662595748901,
      "learning_rate": 6.873734676312835e-06,
      "loss": 0.4776,
      "step": 2639
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.0358595848083496,
      "learning_rate": 6.871479029649489e-06,
      "loss": 0.4913,
      "step": 2640
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3548041582107544,
      "learning_rate": 6.86922293996635e-06,
      "loss": 0.4282,
      "step": 2641
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2319684028625488,
      "learning_rate": 6.866966407797488e-06,
      "loss": 0.391,
      "step": 2642
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3759331703186035,
      "learning_rate": 6.864709433677066e-06,
      "loss": 0.4323,
      "step": 2643
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4887109994888306,
      "learning_rate": 6.862452018139364e-06,
      "loss": 0.5507,
      "step": 2644
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3083797693252563,
      "learning_rate": 6.860194161718759e-06,
      "loss": 0.452,
      "step": 2645
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9864258766174316,
      "learning_rate": 6.857935864949736e-06,
      "loss": 0.5098,
      "step": 2646
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2085567712783813,
      "learning_rate": 6.855677128366881e-06,
      "loss": 0.6997,
      "step": 2647
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5157829523086548,
      "learning_rate": 6.8534179525048895e-06,
      "loss": 0.4784,
      "step": 2648
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1481910943984985,
      "learning_rate": 6.851158337898554e-06,
      "loss": 0.7097,
      "step": 2649
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7967114448547363,
      "learning_rate": 6.8488982850827765e-06,
      "loss": 0.4524,
      "step": 2650
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5961908102035522,
      "learning_rate": 6.84663779459256e-06,
      "loss": 0.44,
      "step": 2651
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.078904390335083,
      "learning_rate": 6.844376866963012e-06,
      "loss": 0.4856,
      "step": 2652
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9380451440811157,
      "learning_rate": 6.842115502729343e-06,
      "loss": 0.4161,
      "step": 2653
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6394398212432861,
      "learning_rate": 6.839853702426868e-06,
      "loss": 0.4704,
      "step": 2654
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6393235921859741,
      "learning_rate": 6.8375914665910026e-06,
      "loss": 0.431,
      "step": 2655
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6037153005599976,
      "learning_rate": 6.835328795757268e-06,
      "loss": 0.4753,
      "step": 2656
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1029080152511597,
      "learning_rate": 6.833065690461289e-06,
      "loss": 0.6903,
      "step": 2657
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.203439235687256,
      "learning_rate": 6.830802151238789e-06,
      "loss": 0.4376,
      "step": 2658
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.677216053009033,
      "learning_rate": 6.828538178625599e-06,
      "loss": 0.4544,
      "step": 2659
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3655143976211548,
      "learning_rate": 6.826273773157652e-06,
      "loss": 0.4515,
      "step": 2660
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.398023843765259,
      "learning_rate": 6.8240089353709785e-06,
      "loss": 0.4505,
      "step": 2661
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5268125534057617,
      "learning_rate": 6.821743665801716e-06,
      "loss": 0.5231,
      "step": 2662
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.124356985092163,
      "learning_rate": 6.819477964986105e-06,
      "loss": 0.5548,
      "step": 2663
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3445804119110107,
      "learning_rate": 6.817211833460484e-06,
      "loss": 0.4187,
      "step": 2664
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2883042097091675,
      "learning_rate": 6.8149452717612965e-06,
      "loss": 0.4674,
      "step": 2665
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3430087566375732,
      "learning_rate": 6.812678280425088e-06,
      "loss": 0.4386,
      "step": 2666
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.298377513885498,
      "learning_rate": 6.8104108599885024e-06,
      "loss": 0.5053,
      "step": 2667
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3108584880828857,
      "learning_rate": 6.8081430109882884e-06,
      "loss": 0.4721,
      "step": 2668
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6145192384719849,
      "learning_rate": 6.805874733961295e-06,
      "loss": 0.4658,
      "step": 2669
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6567562818527222,
      "learning_rate": 6.803606029444474e-06,
      "loss": 0.4977,
      "step": 2670
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1087843179702759,
      "learning_rate": 6.801336897974876e-06,
      "loss": 0.3789,
      "step": 2671
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.601704478263855,
      "learning_rate": 6.7990673400896546e-06,
      "loss": 0.5052,
      "step": 2672
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5529805421829224,
      "learning_rate": 6.796797356326063e-06,
      "loss": 0.4375,
      "step": 2673
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4033455848693848,
      "learning_rate": 6.794526947221454e-06,
      "loss": 0.4127,
      "step": 2674
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.3082170486450195,
      "learning_rate": 6.792256113313283e-06,
      "loss": 0.5274,
      "step": 2675
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6191898584365845,
      "learning_rate": 6.78998485513911e-06,
      "loss": 0.4846,
      "step": 2676
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7620978355407715,
      "learning_rate": 6.787713173236588e-06,
      "loss": 0.4696,
      "step": 2677
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3591561317443848,
      "learning_rate": 6.7854410681434744e-06,
      "loss": 0.3973,
      "step": 2678
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3039767742156982,
      "learning_rate": 6.783168540397625e-06,
      "loss": 0.5434,
      "step": 2679
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5361629724502563,
      "learning_rate": 6.780895590536995e-06,
      "loss": 0.4592,
      "step": 2680
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.464916706085205,
      "learning_rate": 6.778622219099644e-06,
      "loss": 0.4735,
      "step": 2681
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3362698554992676,
      "learning_rate": 6.776348426623729e-06,
      "loss": 0.5556,
      "step": 2682
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.159346342086792,
      "learning_rate": 6.774074213647504e-06,
      "loss": 0.4416,
      "step": 2683
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0867679119110107,
      "learning_rate": 6.771799580709324e-06,
      "loss": 0.7313,
      "step": 2684
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9017378091812134,
      "learning_rate": 6.769524528347647e-06,
      "loss": 0.3982,
      "step": 2685
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1745195388793945,
      "learning_rate": 6.767249057101024e-06,
      "loss": 0.4443,
      "step": 2686
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5760234594345093,
      "learning_rate": 6.764973167508112e-06,
      "loss": 0.5377,
      "step": 2687
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3630291223526,
      "learning_rate": 6.762696860107662e-06,
      "loss": 0.4667,
      "step": 2688
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1951740980148315,
      "learning_rate": 6.760420135438524e-06,
      "loss": 0.4609,
      "step": 2689
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.135908842086792,
      "learning_rate": 6.75814299403965e-06,
      "loss": 0.5184,
      "step": 2690
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.7685937881469727,
      "learning_rate": 6.755865436450088e-06,
      "loss": 0.4943,
      "step": 2691
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1573861837387085,
      "learning_rate": 6.753587463208987e-06,
      "loss": 0.4283,
      "step": 2692
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.1858749389648438,
      "learning_rate": 6.751309074855592e-06,
      "loss": 0.4674,
      "step": 2693
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4373087882995605,
      "learning_rate": 6.7490302719292454e-06,
      "loss": 0.4088,
      "step": 2694
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.2426576614379883,
      "learning_rate": 6.746751054969391e-06,
      "loss": 0.4481,
      "step": 2695
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8487093448638916,
      "learning_rate": 6.74447142451557e-06,
      "loss": 0.5348,
      "step": 2696
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5795818567276,
      "learning_rate": 6.7421913811074194e-06,
      "loss": 0.4443,
      "step": 2697
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.2477262020111084,
      "learning_rate": 6.7399109252846736e-06,
      "loss": 0.4486,
      "step": 2698
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8977912664413452,
      "learning_rate": 6.737630057587169e-06,
      "loss": 0.4758,
      "step": 2699
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4615144729614258,
      "learning_rate": 6.735348778554835e-06,
      "loss": 0.4333,
      "step": 2700
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.963516116142273,
      "learning_rate": 6.733067088727699e-06,
      "loss": 0.4729,
      "step": 2701
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2087509632110596,
      "learning_rate": 6.730784988645889e-06,
      "loss": 0.735,
      "step": 2702
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.752906084060669,
      "learning_rate": 6.728502478849625e-06,
      "loss": 0.5253,
      "step": 2703
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.357031226158142,
      "learning_rate": 6.726219559879229e-06,
      "loss": 0.5239,
      "step": 2704
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8049973249435425,
      "learning_rate": 6.7239362322751166e-06,
      "loss": 0.4862,
      "step": 2705
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6012834310531616,
      "learning_rate": 6.721652496577801e-06,
      "loss": 0.4983,
      "step": 2706
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.2213146686553955,
      "learning_rate": 6.719368353327892e-06,
      "loss": 0.4886,
      "step": 2707
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.014253854751587,
      "learning_rate": 6.717083803066097e-06,
      "loss": 0.4982,
      "step": 2708
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.8772802352905273,
      "learning_rate": 6.7147988463332184e-06,
      "loss": 0.4413,
      "step": 2709
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6015297174453735,
      "learning_rate": 6.712513483670153e-06,
      "loss": 0.471,
      "step": 2710
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4712090492248535,
      "learning_rate": 6.7102277156179e-06,
      "loss": 0.4887,
      "step": 2711
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.2109262943267822,
      "learning_rate": 6.707941542717547e-06,
      "loss": 0.5102,
      "step": 2712
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3756427764892578,
      "learning_rate": 6.705654965510282e-06,
      "loss": 0.524,
      "step": 2713
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1938217878341675,
      "learning_rate": 6.703367984537388e-06,
      "loss": 0.4553,
      "step": 2714
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.25114369392395,
      "learning_rate": 6.701080600340242e-06,
      "loss": 0.4637,
      "step": 2715
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.769222617149353,
      "learning_rate": 6.698792813460318e-06,
      "loss": 0.4958,
      "step": 2716
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.5395612716674805,
      "learning_rate": 6.696504624439185e-06,
      "loss": 0.4368,
      "step": 2717
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2551476955413818,
      "learning_rate": 6.694216033818508e-06,
      "loss": 0.4562,
      "step": 2718
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5534306764602661,
      "learning_rate": 6.691927042140044e-06,
      "loss": 0.5238,
      "step": 2719
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6269499063491821,
      "learning_rate": 6.68963764994565e-06,
      "loss": 0.4871,
      "step": 2720
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4009180068969727,
      "learning_rate": 6.6873478577772745e-06,
      "loss": 0.4117,
      "step": 2721
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2284013032913208,
      "learning_rate": 6.685057666176958e-06,
      "loss": 0.5076,
      "step": 2722
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7462372779846191,
      "learning_rate": 6.682767075686842e-06,
      "loss": 0.459,
      "step": 2723
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2289113998413086,
      "learning_rate": 6.680476086849157e-06,
      "loss": 0.4379,
      "step": 2724
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.533949851989746,
      "learning_rate": 6.67818470020623e-06,
      "loss": 0.4588,
      "step": 2725
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1469969749450684,
      "learning_rate": 6.675892916300486e-06,
      "loss": 0.4222,
      "step": 2726
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1767578125,
      "learning_rate": 6.6736007356744345e-06,
      "loss": 0.7578,
      "step": 2727
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.681768536567688,
      "learning_rate": 6.671308158870687e-06,
      "loss": 0.4888,
      "step": 2728
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5086954832077026,
      "learning_rate": 6.669015186431946e-06,
      "loss": 0.5232,
      "step": 2729
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.623858094215393,
      "learning_rate": 6.666721818901009e-06,
      "loss": 0.4559,
      "step": 2730
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7155518531799316,
      "learning_rate": 6.664428056820764e-06,
      "loss": 0.5081,
      "step": 2731
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3148959875106812,
      "learning_rate": 6.662133900734197e-06,
      "loss": 0.4385,
      "step": 2732
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4000364542007446,
      "learning_rate": 6.6598393511843824e-06,
      "loss": 0.465,
      "step": 2733
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.1721551418304443,
      "learning_rate": 6.657544408714491e-06,
      "loss": 0.4931,
      "step": 2734
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1987475156784058,
      "learning_rate": 6.655249073867784e-06,
      "loss": 0.4845,
      "step": 2735
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7466682195663452,
      "learning_rate": 6.652953347187619e-06,
      "loss": 0.447,
      "step": 2736
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2225383520126343,
      "learning_rate": 6.650657229217445e-06,
      "loss": 0.4443,
      "step": 2737
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3776301145553589,
      "learning_rate": 6.648360720500802e-06,
      "loss": 0.4901,
      "step": 2738
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6846197843551636,
      "learning_rate": 6.646063821581322e-06,
      "loss": 0.5235,
      "step": 2739
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4413647651672363,
      "learning_rate": 6.643766533002734e-06,
      "loss": 0.4368,
      "step": 2740
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1068376302719116,
      "learning_rate": 6.641468855308853e-06,
      "loss": 0.4613,
      "step": 2741
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6117600202560425,
      "learning_rate": 6.63917078904359e-06,
      "loss": 0.519,
      "step": 2742
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4640358686447144,
      "learning_rate": 6.63687233475095e-06,
      "loss": 0.4173,
      "step": 2743
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.3571207523345947,
      "learning_rate": 6.634573492975026e-06,
      "loss": 0.5636,
      "step": 2744
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4980955123901367,
      "learning_rate": 6.6322742642600025e-06,
      "loss": 0.5101,
      "step": 2745
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.735332727432251,
      "learning_rate": 6.629974649150157e-06,
      "loss": 0.4308,
      "step": 2746
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9943512678146362,
      "learning_rate": 6.627674648189858e-06,
      "loss": 0.5504,
      "step": 2747
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2488309144973755,
      "learning_rate": 6.625374261923567e-06,
      "loss": 0.4585,
      "step": 2748
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2804545164108276,
      "learning_rate": 6.623073490895837e-06,
      "loss": 0.4668,
      "step": 2749
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7503876686096191,
      "learning_rate": 6.6207723356513085e-06,
      "loss": 0.4864,
      "step": 2750
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3052273988723755,
      "learning_rate": 6.618470796734714e-06,
      "loss": 0.5274,
      "step": 2751
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4174001216888428,
      "learning_rate": 6.6161688746908805e-06,
      "loss": 0.4507,
      "step": 2752
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9275776147842407,
      "learning_rate": 6.613866570064718e-06,
      "loss": 0.4832,
      "step": 2753
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5921869277954102,
      "learning_rate": 6.611563883401239e-06,
      "loss": 0.5031,
      "step": 2754
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.183607578277588,
      "learning_rate": 6.609260815245535e-06,
      "loss": 0.4548,
      "step": 2755
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2785110473632812,
      "learning_rate": 6.606957366142795e-06,
      "loss": 0.4867,
      "step": 2756
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5207087993621826,
      "learning_rate": 6.604653536638293e-06,
      "loss": 0.4364,
      "step": 2757
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.36026132106781,
      "learning_rate": 6.602349327277397e-06,
      "loss": 0.481,
      "step": 2758
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5448009967803955,
      "learning_rate": 6.600044738605562e-06,
      "loss": 0.5145,
      "step": 2759
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4135586023330688,
      "learning_rate": 6.597739771168336e-06,
      "loss": 0.5223,
      "step": 2760
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4138106107711792,
      "learning_rate": 6.595434425511353e-06,
      "loss": 0.5191,
      "step": 2761
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5271403789520264,
      "learning_rate": 6.5931287021803415e-06,
      "loss": 0.4787,
      "step": 2762
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.469405174255371,
      "learning_rate": 6.590822601721115e-06,
      "loss": 0.5117,
      "step": 2763
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.144634246826172,
      "learning_rate": 6.588516124679575e-06,
      "loss": 0.4884,
      "step": 2764
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5103466510772705,
      "learning_rate": 6.586209271601718e-06,
      "loss": 0.5511,
      "step": 2765
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5000752210617065,
      "learning_rate": 6.583902043033625e-06,
      "loss": 0.4967,
      "step": 2766
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.730250358581543,
      "learning_rate": 6.581594439521468e-06,
      "loss": 0.5228,
      "step": 2767
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5698906183242798,
      "learning_rate": 6.579286461611506e-06,
      "loss": 0.4626,
      "step": 2768
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6053334474563599,
      "learning_rate": 6.576978109850088e-06,
      "loss": 0.4434,
      "step": 2769
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2934099435806274,
      "learning_rate": 6.574669384783652e-06,
      "loss": 0.4295,
      "step": 2770
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1561973094940186,
      "learning_rate": 6.572360286958721e-06,
      "loss": 0.404,
      "step": 2771
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.306473731994629,
      "learning_rate": 6.570050816921911e-06,
      "loss": 0.4478,
      "step": 2772
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.4771876335144043,
      "learning_rate": 6.567740975219923e-06,
      "loss": 0.4906,
      "step": 2773
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5907529592514038,
      "learning_rate": 6.565430762399546e-06,
      "loss": 0.4468,
      "step": 2774
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4764704704284668,
      "learning_rate": 6.563120179007659e-06,
      "loss": 0.5084,
      "step": 2775
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6144981384277344,
      "learning_rate": 6.560809225591227e-06,
      "loss": 0.4769,
      "step": 2776
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7248326539993286,
      "learning_rate": 6.5584979026973014e-06,
      "loss": 0.461,
      "step": 2777
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4106196165084839,
      "learning_rate": 6.556186210873025e-06,
      "loss": 0.4747,
      "step": 2778
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3201969861984253,
      "learning_rate": 6.5538741506656236e-06,
      "loss": 0.5558,
      "step": 2779
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4280422925949097,
      "learning_rate": 6.551561722622414e-06,
      "loss": 0.4551,
      "step": 2780
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.2470178604125977,
      "learning_rate": 6.549248927290796e-06,
      "loss": 0.4598,
      "step": 2781
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1620807647705078,
      "learning_rate": 6.54693576521826e-06,
      "loss": 0.4937,
      "step": 2782
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9011634588241577,
      "learning_rate": 6.544622236952382e-06,
      "loss": 0.4247,
      "step": 2783
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4119716882705688,
      "learning_rate": 6.542308343040823e-06,
      "loss": 0.4789,
      "step": 2784
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.731943130493164,
      "learning_rate": 6.539994084031334e-06,
      "loss": 0.5412,
      "step": 2785
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4092566967010498,
      "learning_rate": 6.537679460471747e-06,
      "loss": 0.4653,
      "step": 2786
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5789072513580322,
      "learning_rate": 6.535364472909989e-06,
      "loss": 0.434,
      "step": 2787
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3377054929733276,
      "learning_rate": 6.533049121894064e-06,
      "loss": 0.5058,
      "step": 2788
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3771427869796753,
      "learning_rate": 6.530733407972067e-06,
      "loss": 0.5045,
      "step": 2789
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.1772708892822266,
      "learning_rate": 6.528417331692177e-06,
      "loss": 0.5697,
      "step": 2790
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.561937689781189,
      "learning_rate": 6.526100893602661e-06,
      "loss": 0.4782,
      "step": 2791
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.085742235183716,
      "learning_rate": 6.5237840942518706e-06,
      "loss": 0.5362,
      "step": 2792
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.180056571960449,
      "learning_rate": 6.521466934188241e-06,
      "loss": 0.5159,
      "step": 2793
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.029573440551758,
      "learning_rate": 6.519149413960295e-06,
      "loss": 0.5244,
      "step": 2794
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7928581237792969,
      "learning_rate": 6.51683153411664e-06,
      "loss": 0.4974,
      "step": 2795
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.7994160652160645,
      "learning_rate": 6.51451329520597e-06,
      "loss": 0.4447,
      "step": 2796
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.875367522239685,
      "learning_rate": 6.5121946977770585e-06,
      "loss": 0.4589,
      "step": 2797
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.2139363288879395,
      "learning_rate": 6.509875742378774e-06,
      "loss": 0.4756,
      "step": 2798
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8431787490844727,
      "learning_rate": 6.5075564295600594e-06,
      "loss": 0.4311,
      "step": 2799
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.0177321434020996,
      "learning_rate": 6.505236759869947e-06,
      "loss": 0.498,
      "step": 2800
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.91667103767395,
      "learning_rate": 6.502916733857554e-06,
      "loss": 0.5601,
      "step": 2801
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.121142029762268,
      "learning_rate": 6.50059635207208e-06,
      "loss": 0.7441,
      "step": 2802
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.604210376739502,
      "learning_rate": 6.49827561506281e-06,
      "loss": 0.4708,
      "step": 2803
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6532206535339355,
      "learning_rate": 6.4959545233791155e-06,
      "loss": 0.4875,
      "step": 2804
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.023980140686035,
      "learning_rate": 6.493633077570446e-06,
      "loss": 0.5503,
      "step": 2805
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.0794405937194824,
      "learning_rate": 6.491311278186339e-06,
      "loss": 0.5108,
      "step": 2806
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9052990674972534,
      "learning_rate": 6.488989125776417e-06,
      "loss": 0.5246,
      "step": 2807
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9026299715042114,
      "learning_rate": 6.48666662089038e-06,
      "loss": 0.5228,
      "step": 2808
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.376488208770752,
      "learning_rate": 6.484343764078019e-06,
      "loss": 0.3832,
      "step": 2809
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9965360164642334,
      "learning_rate": 6.482020555889206e-06,
      "loss": 0.4536,
      "step": 2810
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1713577508926392,
      "learning_rate": 6.4796969968738895e-06,
      "loss": 0.6718,
      "step": 2811
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5963574647903442,
      "learning_rate": 6.47737308758211e-06,
      "loss": 0.4206,
      "step": 2812
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4951306581497192,
      "learning_rate": 6.475048828563988e-06,
      "loss": 0.4974,
      "step": 2813
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.624263048171997,
      "learning_rate": 6.4727242203697215e-06,
      "loss": 0.4768,
      "step": 2814
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3454372882843018,
      "learning_rate": 6.470399263549603e-06,
      "loss": 0.4527,
      "step": 2815
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4755055904388428,
      "learning_rate": 6.4680739586539956e-06,
      "loss": 0.5376,
      "step": 2816
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7254480123519897,
      "learning_rate": 6.46574830623335e-06,
      "loss": 0.4707,
      "step": 2817
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8706337213516235,
      "learning_rate": 6.463422306838199e-06,
      "loss": 0.4795,
      "step": 2818
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.492397427558899,
      "learning_rate": 6.461095961019158e-06,
      "loss": 0.4459,
      "step": 2819
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.431570529937744,
      "learning_rate": 6.4587692693269235e-06,
      "loss": 0.4965,
      "step": 2820
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.645473599433899,
      "learning_rate": 6.4564422323122735e-06,
      "loss": 0.5015,
      "step": 2821
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7296983003616333,
      "learning_rate": 6.45411485052607e-06,
      "loss": 0.4523,
      "step": 2822
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1241310834884644,
      "learning_rate": 6.451787124519251e-06,
      "loss": 0.425,
      "step": 2823
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9997754096984863,
      "learning_rate": 6.449459054842844e-06,
      "loss": 0.4855,
      "step": 2824
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6275020837783813,
      "learning_rate": 6.447130642047949e-06,
      "loss": 0.4533,
      "step": 2825
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9246485233306885,
      "learning_rate": 6.444801886685758e-06,
      "loss": 0.439,
      "step": 2826
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.120377779006958,
      "learning_rate": 6.442472789307533e-06,
      "loss": 0.4509,
      "step": 2827
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9024643898010254,
      "learning_rate": 6.440143350464625e-06,
      "loss": 0.459,
      "step": 2828
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.646507740020752,
      "learning_rate": 6.437813570708463e-06,
      "loss": 0.4801,
      "step": 2829
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1872162818908691,
      "learning_rate": 6.435483450590554e-06,
      "loss": 0.4434,
      "step": 2830
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5768458843231201,
      "learning_rate": 6.433152990662489e-06,
      "loss": 0.5199,
      "step": 2831
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.515557885169983,
      "learning_rate": 6.43082219147594e-06,
      "loss": 0.561,
      "step": 2832
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9241523742675781,
      "learning_rate": 6.428491053582657e-06,
      "loss": 0.4907,
      "step": 2833
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.460750102996826,
      "learning_rate": 6.42615957753447e-06,
      "loss": 0.4591,
      "step": 2834
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.7661478519439697,
      "learning_rate": 6.423827763883294e-06,
      "loss": 0.4572,
      "step": 2835
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8398698568344116,
      "learning_rate": 6.421495613181112e-06,
      "loss": 0.4682,
      "step": 2836
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6801655292510986,
      "learning_rate": 6.419163125980004e-06,
      "loss": 0.5314,
      "step": 2837
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5028846263885498,
      "learning_rate": 6.416830302832114e-06,
      "loss": 0.4758,
      "step": 2838
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6711809635162354,
      "learning_rate": 6.414497144289675e-06,
      "loss": 0.5018,
      "step": 2839
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4078640937805176,
      "learning_rate": 6.4121636509049965e-06,
      "loss": 0.5262,
      "step": 2840
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8074883222579956,
      "learning_rate": 6.409829823230467e-06,
      "loss": 0.4615,
      "step": 2841
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.201611042022705,
      "learning_rate": 6.407495661818551e-06,
      "loss": 0.5467,
      "step": 2842
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.1082937717437744,
      "learning_rate": 6.405161167221799e-06,
      "loss": 0.4558,
      "step": 2843
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6400527954101562,
      "learning_rate": 6.402826339992836e-06,
      "loss": 0.4826,
      "step": 2844
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7862154245376587,
      "learning_rate": 6.400491180684365e-06,
      "loss": 0.4944,
      "step": 2845
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7259217500686646,
      "learning_rate": 6.3981556898491705e-06,
      "loss": 0.5146,
      "step": 2846
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.136582374572754,
      "learning_rate": 6.395819868040114e-06,
      "loss": 0.6997,
      "step": 2847
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4497363567352295,
      "learning_rate": 6.393483715810134e-06,
      "loss": 0.4808,
      "step": 2848
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6182425022125244,
      "learning_rate": 6.391147233712248e-06,
      "loss": 0.5683,
      "step": 2849
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4409878253936768,
      "learning_rate": 6.388810422299555e-06,
      "loss": 0.4481,
      "step": 2850
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5567946434020996,
      "learning_rate": 6.386473282125227e-06,
      "loss": 0.5668,
      "step": 2851
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5510839223861694,
      "learning_rate": 6.384135813742516e-06,
      "loss": 0.5369,
      "step": 2852
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.644863247871399,
      "learning_rate": 6.381798017704752e-06,
      "loss": 0.4655,
      "step": 2853
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4384040832519531,
      "learning_rate": 6.379459894565343e-06,
      "loss": 0.4671,
      "step": 2854
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9065721035003662,
      "learning_rate": 6.377121444877773e-06,
      "loss": 0.49,
      "step": 2855
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.888877034187317,
      "learning_rate": 6.374782669195603e-06,
      "loss": 0.4364,
      "step": 2856
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5239136219024658,
      "learning_rate": 6.372443568072473e-06,
      "loss": 0.4567,
      "step": 2857
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5374305248260498,
      "learning_rate": 6.370104142062098e-06,
      "loss": 0.4614,
      "step": 2858
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.150045156478882,
      "learning_rate": 6.367764391718272e-06,
      "loss": 0.4871,
      "step": 2859
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.710041046142578,
      "learning_rate": 6.3654243175948646e-06,
      "loss": 0.5139,
      "step": 2860
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.4196059703826904,
      "learning_rate": 6.363083920245824e-06,
      "loss": 0.4632,
      "step": 2861
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.124279499053955,
      "learning_rate": 6.36074320022517e-06,
      "loss": 0.5299,
      "step": 2862
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.6314995288848877,
      "learning_rate": 6.358402158087003e-06,
      "loss": 0.4401,
      "step": 2863
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.4334487915039062,
      "learning_rate": 6.3560607943855e-06,
      "loss": 0.484,
      "step": 2864
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7813739776611328,
      "learning_rate": 6.353719109674912e-06,
      "loss": 0.4662,
      "step": 2865
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.357008934020996,
      "learning_rate": 6.3513771045095655e-06,
      "loss": 0.4696,
      "step": 2866
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8631997108459473,
      "learning_rate": 6.349034779443865e-06,
      "loss": 0.5727,
      "step": 2867
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4755865335464478,
      "learning_rate": 6.3466921350322894e-06,
      "loss": 0.5245,
      "step": 2868
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5736562013626099,
      "learning_rate": 6.3443491718293945e-06,
      "loss": 0.4597,
      "step": 2869
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9663130044937134,
      "learning_rate": 6.342005890389811e-06,
      "loss": 0.513,
      "step": 2870
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5723803043365479,
      "learning_rate": 6.339662291268243e-06,
      "loss": 0.4912,
      "step": 2871
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5614174604415894,
      "learning_rate": 6.337318375019472e-06,
      "loss": 0.4343,
      "step": 2872
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8432800769805908,
      "learning_rate": 6.334974142198354e-06,
      "loss": 0.4926,
      "step": 2873
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.098661422729492,
      "learning_rate": 6.332629593359821e-06,
      "loss": 0.5202,
      "step": 2874
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9976519346237183,
      "learning_rate": 6.330284729058874e-06,
      "loss": 0.4722,
      "step": 2875
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.677011013031006,
      "learning_rate": 6.327939549850602e-06,
      "loss": 0.4441,
      "step": 2876
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.89258873462677,
      "learning_rate": 6.325594056290152e-06,
      "loss": 0.4402,
      "step": 2877
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1540347337722778,
      "learning_rate": 6.3232482489327565e-06,
      "loss": 0.7268,
      "step": 2878
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5045201778411865,
      "learning_rate": 6.320902128333719e-06,
      "loss": 0.4629,
      "step": 2879
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7581154108047485,
      "learning_rate": 6.318555695048418e-06,
      "loss": 0.4814,
      "step": 2880
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.5720620155334473,
      "learning_rate": 6.316208949632303e-06,
      "loss": 0.4651,
      "step": 2881
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.611686110496521,
      "learning_rate": 6.3138618926409025e-06,
      "loss": 0.4214,
      "step": 2882
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8348345756530762,
      "learning_rate": 6.311514524629815e-06,
      "loss": 0.5556,
      "step": 2883
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.476425290107727,
      "learning_rate": 6.309166846154713e-06,
      "loss": 0.4744,
      "step": 2884
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2976531982421875,
      "learning_rate": 6.306818857771342e-06,
      "loss": 0.4254,
      "step": 2885
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3154247999191284,
      "learning_rate": 6.304470560035522e-06,
      "loss": 0.4015,
      "step": 2886
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.880855917930603,
      "learning_rate": 6.30212195350315e-06,
      "loss": 0.4669,
      "step": 2887
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2986842393875122,
      "learning_rate": 6.299773038730189e-06,
      "loss": 0.776,
      "step": 2888
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5989625453948975,
      "learning_rate": 6.297423816272679e-06,
      "loss": 0.534,
      "step": 2889
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6473684310913086,
      "learning_rate": 6.295074286686731e-06,
      "loss": 0.4857,
      "step": 2890
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.5443477630615234,
      "learning_rate": 6.292724450528531e-06,
      "loss": 0.4287,
      "step": 2891
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.6794731616973877,
      "learning_rate": 6.2903743083543346e-06,
      "loss": 0.5507,
      "step": 2892
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5869953632354736,
      "learning_rate": 6.288023860720475e-06,
      "loss": 0.4395,
      "step": 2893
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8200584650039673,
      "learning_rate": 6.285673108183352e-06,
      "loss": 0.462,
      "step": 2894
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.920008897781372,
      "learning_rate": 6.283322051299439e-06,
      "loss": 0.4507,
      "step": 2895
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6790739297866821,
      "learning_rate": 6.280970690625283e-06,
      "loss": 0.4968,
      "step": 2896
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4484857320785522,
      "learning_rate": 6.278619026717503e-06,
      "loss": 0.4635,
      "step": 2897
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.2193727493286133,
      "learning_rate": 6.2762670601327875e-06,
      "loss": 0.5227,
      "step": 2898
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.216361403465271,
      "learning_rate": 6.273914791427902e-06,
      "loss": 0.4703,
      "step": 2899
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7848228216171265,
      "learning_rate": 6.271562221159676e-06,
      "loss": 0.5041,
      "step": 2900
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1524505615234375,
      "learning_rate": 6.269209349885014e-06,
      "loss": 0.3969,
      "step": 2901
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3480814695358276,
      "learning_rate": 6.266856178160892e-06,
      "loss": 0.4771,
      "step": 2902
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9322913885116577,
      "learning_rate": 6.264502706544358e-06,
      "loss": 0.4471,
      "step": 2903
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5282807350158691,
      "learning_rate": 6.26214893559253e-06,
      "loss": 0.4694,
      "step": 2904
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3897483348846436,
      "learning_rate": 6.259794865862596e-06,
      "loss": 0.4371,
      "step": 2905
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9069430828094482,
      "learning_rate": 6.257440497911816e-06,
      "loss": 0.496,
      "step": 2906
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.695562720298767,
      "learning_rate": 6.255085832297522e-06,
      "loss": 0.5407,
      "step": 2907
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.259128451347351,
      "learning_rate": 6.25273086957711e-06,
      "loss": 0.4879,
      "step": 2908
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.394066572189331,
      "learning_rate": 6.250375610308054e-06,
      "loss": 0.4218,
      "step": 2909
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3879717588424683,
      "learning_rate": 6.248020055047894e-06,
      "loss": 0.4757,
      "step": 2910
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9846030473709106,
      "learning_rate": 6.245664204354242e-06,
      "loss": 0.4983,
      "step": 2911
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8236602544784546,
      "learning_rate": 6.24330805878478e-06,
      "loss": 0.5147,
      "step": 2912
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.2004213333129883,
      "learning_rate": 6.2409516188972574e-06,
      "loss": 0.4788,
      "step": 2913
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8153287172317505,
      "learning_rate": 6.238594885249493e-06,
      "loss": 0.5164,
      "step": 2914
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5611387491226196,
      "learning_rate": 6.236237858399381e-06,
      "loss": 0.5124,
      "step": 2915
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7838677167892456,
      "learning_rate": 6.233880538904878e-06,
      "loss": 0.4552,
      "step": 2916
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.578039526939392,
      "learning_rate": 6.231522927324014e-06,
      "loss": 0.4943,
      "step": 2917
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2379770278930664,
      "learning_rate": 6.229165024214886e-06,
      "loss": 0.4502,
      "step": 2918
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.249300956726074,
      "learning_rate": 6.226806830135663e-06,
      "loss": 0.4582,
      "step": 2919
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8095340728759766,
      "learning_rate": 6.224448345644578e-06,
      "loss": 0.5273,
      "step": 2920
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6361448764801025,
      "learning_rate": 6.222089571299937e-06,
      "loss": 0.5127,
      "step": 2921
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5533781051635742,
      "learning_rate": 6.2197305076601145e-06,
      "loss": 0.5154,
      "step": 2922
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1960049867630005,
      "learning_rate": 6.217371155283551e-06,
      "loss": 0.3659,
      "step": 2923
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6041347980499268,
      "learning_rate": 6.215011514728755e-06,
      "loss": 0.4384,
      "step": 2924
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4405308961868286,
      "learning_rate": 6.212651586554306e-06,
      "loss": 0.4242,
      "step": 2925
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0962049961090088,
      "learning_rate": 6.210291371318852e-06,
      "loss": 0.719,
      "step": 2926
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.928870439529419,
      "learning_rate": 6.207930869581105e-06,
      "loss": 0.4784,
      "step": 2927
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.3174493312835693,
      "learning_rate": 6.205570081899847e-06,
      "loss": 0.5498,
      "step": 2928
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4363203048706055,
      "learning_rate": 6.203209008833929e-06,
      "loss": 0.4608,
      "step": 2929
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4593300819396973,
      "learning_rate": 6.200847650942269e-06,
      "loss": 0.4756,
      "step": 2930
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4942978620529175,
      "learning_rate": 6.1984860087838495e-06,
      "loss": 0.474,
      "step": 2931
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0866605043411255,
      "learning_rate": 6.196124082917725e-06,
      "loss": 0.5246,
      "step": 2932
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3494982719421387,
      "learning_rate": 6.1937618739030125e-06,
      "loss": 0.4855,
      "step": 2933
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.5813400745391846,
      "learning_rate": 6.1913993822988995e-06,
      "loss": 0.4929,
      "step": 2934
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0787010192871094,
      "learning_rate": 6.189036608664639e-06,
      "loss": 0.4708,
      "step": 2935
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6219935417175293,
      "learning_rate": 6.186673553559551e-06,
      "loss": 0.441,
      "step": 2936
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.506768822669983,
      "learning_rate": 6.184310217543022e-06,
      "loss": 0.4625,
      "step": 2937
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.710073471069336,
      "learning_rate": 6.181946601174506e-06,
      "loss": 0.4709,
      "step": 2938
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7805284261703491,
      "learning_rate": 6.179582705013519e-06,
      "loss": 0.4341,
      "step": 2939
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3899725675582886,
      "learning_rate": 6.177218529619651e-06,
      "loss": 0.5268,
      "step": 2940
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.320043683052063,
      "learning_rate": 6.174854075552551e-06,
      "loss": 0.4653,
      "step": 2941
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.406440019607544,
      "learning_rate": 6.172489343371937e-06,
      "loss": 0.4878,
      "step": 2942
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.280396819114685,
      "learning_rate": 6.1701243336375935e-06,
      "loss": 0.3964,
      "step": 2943
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4168568849563599,
      "learning_rate": 6.16775904690937e-06,
      "loss": 0.4307,
      "step": 2944
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0406548976898193,
      "learning_rate": 6.165393483747179e-06,
      "loss": 0.6862,
      "step": 2945
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.267844319343567,
      "learning_rate": 6.163027644711003e-06,
      "loss": 0.5104,
      "step": 2946
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4223850965499878,
      "learning_rate": 6.160661530360887e-06,
      "loss": 0.477,
      "step": 2947
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.925194501876831,
      "learning_rate": 6.1582951412569415e-06,
      "loss": 0.5007,
      "step": 2948
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.705609917640686,
      "learning_rate": 6.155928477959342e-06,
      "loss": 0.4908,
      "step": 2949
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.872369647026062,
      "learning_rate": 6.15356154102833e-06,
      "loss": 0.5094,
      "step": 2950
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2256321907043457,
      "learning_rate": 6.15119433102421e-06,
      "loss": 0.7225,
      "step": 2951
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6524925231933594,
      "learning_rate": 6.1488268485073536e-06,
      "loss": 0.4655,
      "step": 2952
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3931602239608765,
      "learning_rate": 6.1464590940381914e-06,
      "loss": 0.4431,
      "step": 2953
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.439328670501709,
      "learning_rate": 6.144091068177229e-06,
      "loss": 0.4406,
      "step": 2954
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2676578760147095,
      "learning_rate": 6.141722771485024e-06,
      "loss": 0.4132,
      "step": 2955
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2814905643463135,
      "learning_rate": 6.139354204522206e-06,
      "loss": 0.4592,
      "step": 2956
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6844359636306763,
      "learning_rate": 6.136985367849467e-06,
      "loss": 0.4999,
      "step": 2957
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.508827805519104,
      "learning_rate": 6.13461626202756e-06,
      "loss": 0.5213,
      "step": 2958
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2647327184677124,
      "learning_rate": 6.132246887617306e-06,
      "loss": 0.441,
      "step": 2959
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3900601863861084,
      "learning_rate": 6.129877245179587e-06,
      "loss": 0.4383,
      "step": 2960
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3924002647399902,
      "learning_rate": 6.127507335275348e-06,
      "loss": 0.4771,
      "step": 2961
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.120217204093933,
      "learning_rate": 6.125137158465598e-06,
      "loss": 0.4157,
      "step": 2962
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7191625833511353,
      "learning_rate": 6.12276671531141e-06,
      "loss": 0.5379,
      "step": 2963
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.362612247467041,
      "learning_rate": 6.1203960063739185e-06,
      "loss": 0.4925,
      "step": 2964
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3560690879821777,
      "learning_rate": 6.1180250322143255e-06,
      "loss": 0.4657,
      "step": 2965
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3814526796340942,
      "learning_rate": 6.11565379339389e-06,
      "loss": 0.4601,
      "step": 2966
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5039546489715576,
      "learning_rate": 6.113282290473933e-06,
      "loss": 0.552,
      "step": 2967
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.772298812866211,
      "learning_rate": 6.110910524015844e-06,
      "loss": 0.4662,
      "step": 2968
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.367432713508606,
      "learning_rate": 6.108538494581071e-06,
      "loss": 0.5703,
      "step": 2969
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.291988730430603,
      "learning_rate": 6.106166202731123e-06,
      "loss": 0.448,
      "step": 2970
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.2445945739746094,
      "learning_rate": 6.103793649027577e-06,
      "loss": 0.4402,
      "step": 2971
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0944007635116577,
      "learning_rate": 6.101420834032067e-06,
      "loss": 0.47,
      "step": 2972
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2092039585113525,
      "learning_rate": 6.099047758306288e-06,
      "loss": 0.5133,
      "step": 2973
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7407830953598022,
      "learning_rate": 6.096674422411999e-06,
      "loss": 0.4825,
      "step": 2974
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2418203353881836,
      "learning_rate": 6.094300826911021e-06,
      "loss": 0.4877,
      "step": 2975
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7367311716079712,
      "learning_rate": 6.091926972365234e-06,
      "loss": 0.4093,
      "step": 2976
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5095524787902832,
      "learning_rate": 6.089552859336585e-06,
      "loss": 0.5063,
      "step": 2977
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3108290433883667,
      "learning_rate": 6.087178488387074e-06,
      "loss": 0.4387,
      "step": 2978
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1329106092453003,
      "learning_rate": 6.08480386007877e-06,
      "loss": 0.4883,
      "step": 2979
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2002872228622437,
      "learning_rate": 6.0824289749737955e-06,
      "loss": 0.5665,
      "step": 2980
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.327897071838379,
      "learning_rate": 6.080053833634338e-06,
      "loss": 0.4634,
      "step": 2981
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3737553358078003,
      "learning_rate": 6.077678436622647e-06,
      "loss": 0.3958,
      "step": 2982
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4687293767929077,
      "learning_rate": 6.07530278450103e-06,
      "loss": 0.4645,
      "step": 2983
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.426309108734131,
      "learning_rate": 6.0729268778318566e-06,
      "loss": 0.4458,
      "step": 2984
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.464289665222168,
      "learning_rate": 6.070550717177553e-06,
      "loss": 0.4433,
      "step": 2985
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2100144624710083,
      "learning_rate": 6.068174303100609e-06,
      "loss": 0.4829,
      "step": 2986
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1961532831192017,
      "learning_rate": 6.065797636163575e-06,
      "loss": 0.4902,
      "step": 2987
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.391446590423584,
      "learning_rate": 6.063420716929058e-06,
      "loss": 0.4871,
      "step": 2988
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1778351068496704,
      "learning_rate": 6.061043545959727e-06,
      "loss": 0.4765,
      "step": 2989
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5456219911575317,
      "learning_rate": 6.058666123818311e-06,
      "loss": 0.4462,
      "step": 2990
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3794018030166626,
      "learning_rate": 6.056288451067598e-06,
      "loss": 0.5108,
      "step": 2991
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.497842788696289,
      "learning_rate": 6.053910528270431e-06,
      "loss": 0.4086,
      "step": 2992
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1665067672729492,
      "learning_rate": 6.05153235598972e-06,
      "loss": 0.4583,
      "step": 2993
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1449270248413086,
      "learning_rate": 6.049153934788429e-06,
      "loss": 0.4873,
      "step": 2994
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2814866304397583,
      "learning_rate": 6.0467752652295805e-06,
      "loss": 0.4417,
      "step": 2995
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.1959919929504395,
      "learning_rate": 6.04439634787626e-06,
      "loss": 0.5636,
      "step": 2996
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0284346342086792,
      "learning_rate": 6.042017183291607e-06,
      "loss": 0.4414,
      "step": 2997
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1874819993972778,
      "learning_rate": 6.039637772038822e-06,
      "loss": 0.5317,
      "step": 2998
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.007399320602417,
      "learning_rate": 6.037258114681164e-06,
      "loss": 0.5154,
      "step": 2999
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.65886390209198,
      "learning_rate": 6.034878211781949e-06,
      "loss": 0.5259,
      "step": 3000
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0874252319335938,
      "learning_rate": 6.032498063904551e-06,
      "loss": 0.4797,
      "step": 3001
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0532433986663818,
      "learning_rate": 6.030117671612405e-06,
      "loss": 0.4786,
      "step": 3002
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1573057174682617,
      "learning_rate": 6.027737035469e-06,
      "loss": 0.4138,
      "step": 3003
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.291890263557434,
      "learning_rate": 6.025356156037885e-06,
      "loss": 0.4717,
      "step": 3004
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3888100385665894,
      "learning_rate": 6.022975033882665e-06,
      "loss": 0.4711,
      "step": 3005
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4528566598892212,
      "learning_rate": 6.020593669567005e-06,
      "loss": 0.4801,
      "step": 3006
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4059001207351685,
      "learning_rate": 6.018212063654625e-06,
      "loss": 0.4476,
      "step": 3007
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4265066385269165,
      "learning_rate": 6.0158302167093016e-06,
      "loss": 0.4719,
      "step": 3008
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6868499517440796,
      "learning_rate": 6.013448129294871e-06,
      "loss": 0.4478,
      "step": 3009
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4533365964889526,
      "learning_rate": 6.011065801975225e-06,
      "loss": 0.5487,
      "step": 3010
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5624018907546997,
      "learning_rate": 6.008683235314313e-06,
      "loss": 0.4703,
      "step": 3011
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4512054920196533,
      "learning_rate": 6.00630042987614e-06,
      "loss": 0.4355,
      "step": 3012
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.1779837608337402,
      "learning_rate": 6.003917386224766e-06,
      "loss": 0.4912,
      "step": 3013
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.034241199493408,
      "learning_rate": 6.001534104924311e-06,
      "loss": 0.4764,
      "step": 3014
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2208653688430786,
      "learning_rate": 5.999150586538951e-06,
      "loss": 0.4068,
      "step": 3015
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.310773491859436,
      "learning_rate": 5.996766831632913e-06,
      "loss": 0.3837,
      "step": 3016
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1787798404693604,
      "learning_rate": 5.994382840770488e-06,
      "loss": 0.4774,
      "step": 3017
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1570266485214233,
      "learning_rate": 5.991998614516014e-06,
      "loss": 0.5152,
      "step": 3018
    },
    {
      "epoch": 0.45,
      "grad_norm": 8.105323791503906,
      "learning_rate": 5.989614153433891e-06,
      "loss": 0.483,
      "step": 3019
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.301736831665039,
      "learning_rate": 5.987229458088573e-06,
      "loss": 0.5443,
      "step": 3020
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.412809133529663,
      "learning_rate": 5.984844529044571e-06,
      "loss": 0.4857,
      "step": 3021
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7968329191207886,
      "learning_rate": 5.982459366866448e-06,
      "loss": 0.5699,
      "step": 3022
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3649382591247559,
      "learning_rate": 5.980073972118822e-06,
      "loss": 0.5092,
      "step": 3023
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3439100980758667,
      "learning_rate": 5.9776883453663705e-06,
      "loss": 0.4583,
      "step": 3024
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7766647338867188,
      "learning_rate": 5.975302487173822e-06,
      "loss": 0.5516,
      "step": 3025
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4629822969436646,
      "learning_rate": 5.972916398105961e-06,
      "loss": 0.4938,
      "step": 3026
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3898066282272339,
      "learning_rate": 5.9705300787276275e-06,
      "loss": 0.5133,
      "step": 3027
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.474007248878479,
      "learning_rate": 5.968143529603715e-06,
      "loss": 0.4948,
      "step": 3028
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1607455015182495,
      "learning_rate": 5.965756751299171e-06,
      "loss": 0.5076,
      "step": 3029
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.160327434539795,
      "learning_rate": 5.963369744378998e-06,
      "loss": 0.4768,
      "step": 3030
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3811488151550293,
      "learning_rate": 5.960982509408252e-06,
      "loss": 0.4681,
      "step": 3031
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.398617148399353,
      "learning_rate": 5.958595046952045e-06,
      "loss": 0.4394,
      "step": 3032
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2331849336624146,
      "learning_rate": 5.9562073575755396e-06,
      "loss": 0.7528,
      "step": 3033
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2254302501678467,
      "learning_rate": 5.953819441843954e-06,
      "loss": 0.4852,
      "step": 3034
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2653921842575073,
      "learning_rate": 5.951431300322562e-06,
      "loss": 0.5068,
      "step": 3035
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5486866235733032,
      "learning_rate": 5.949042933576685e-06,
      "loss": 0.5012,
      "step": 3036
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.394671082496643,
      "learning_rate": 5.946654342171704e-06,
      "loss": 0.5146,
      "step": 3037
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3043681383132935,
      "learning_rate": 5.944265526673051e-06,
      "loss": 0.5051,
      "step": 3038
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3941545486450195,
      "learning_rate": 5.941876487646208e-06,
      "loss": 0.5035,
      "step": 3039
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.275391936302185,
      "learning_rate": 5.939487225656715e-06,
      "loss": 0.4923,
      "step": 3040
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1972296237945557,
      "learning_rate": 5.93709774127016e-06,
      "loss": 0.4301,
      "step": 3041
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3092869520187378,
      "learning_rate": 5.934708035052187e-06,
      "loss": 0.4515,
      "step": 3042
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7733148336410522,
      "learning_rate": 5.932318107568492e-06,
      "loss": 0.4852,
      "step": 3043
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.3106157779693604,
      "learning_rate": 5.929927959384823e-06,
      "loss": 0.4342,
      "step": 3044
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.220184564590454,
      "learning_rate": 5.92753759106698e-06,
      "loss": 0.4942,
      "step": 3045
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3638744354248047,
      "learning_rate": 5.925147003180813e-06,
      "loss": 0.446,
      "step": 3046
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.449945330619812,
      "learning_rate": 5.922756196292228e-06,
      "loss": 0.4353,
      "step": 3047
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3303396701812744,
      "learning_rate": 5.920365170967181e-06,
      "loss": 0.4879,
      "step": 3048
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.807713270187378,
      "learning_rate": 5.917973927771678e-06,
      "loss": 0.3877,
      "step": 3049
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3230518102645874,
      "learning_rate": 5.915582467271782e-06,
      "loss": 0.4802,
      "step": 3050
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2674190998077393,
      "learning_rate": 5.913190790033603e-06,
      "loss": 0.5467,
      "step": 3051
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.132512092590332,
      "learning_rate": 5.910798896623299e-06,
      "loss": 0.457,
      "step": 3052
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4914041757583618,
      "learning_rate": 5.908406787607084e-06,
      "loss": 0.5189,
      "step": 3053
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5744004249572754,
      "learning_rate": 5.9060144635512265e-06,
      "loss": 0.5503,
      "step": 3054
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1754977703094482,
      "learning_rate": 5.9036219250220384e-06,
      "loss": 0.4595,
      "step": 3055
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5650051832199097,
      "learning_rate": 5.901229172585886e-06,
      "loss": 0.4955,
      "step": 3056
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3187172412872314,
      "learning_rate": 5.898836206809188e-06,
      "loss": 0.4406,
      "step": 3057
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1976186037063599,
      "learning_rate": 5.8964430282584075e-06,
      "loss": 0.4951,
      "step": 3058
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.164891004562378,
      "learning_rate": 5.894049637500063e-06,
      "loss": 0.5065,
      "step": 3059
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2375935316085815,
      "learning_rate": 5.891656035100725e-06,
      "loss": 0.573,
      "step": 3060
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2908961772918701,
      "learning_rate": 5.889262221627008e-06,
      "loss": 0.4983,
      "step": 3061
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6564244031906128,
      "learning_rate": 5.886868197645582e-06,
      "loss": 0.5003,
      "step": 3062
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.489245891571045,
      "learning_rate": 5.884473963723164e-06,
      "loss": 0.4749,
      "step": 3063
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.296067476272583,
      "learning_rate": 5.88207952042652e-06,
      "loss": 0.53,
      "step": 3064
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4403644800186157,
      "learning_rate": 5.879684868322468e-06,
      "loss": 0.4351,
      "step": 3065
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2416536808013916,
      "learning_rate": 5.877290007977874e-06,
      "loss": 0.447,
      "step": 3066
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1184762716293335,
      "learning_rate": 5.874894939959653e-06,
      "loss": 0.4301,
      "step": 3067
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.092016577720642,
      "learning_rate": 5.872499664834771e-06,
      "loss": 0.5209,
      "step": 3068
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6310759782791138,
      "learning_rate": 5.870104183170241e-06,
      "loss": 0.4044,
      "step": 3069
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.017486810684204,
      "learning_rate": 5.867708495533124e-06,
      "loss": 0.4206,
      "step": 3070
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8991931676864624,
      "learning_rate": 5.865312602490534e-06,
      "loss": 0.4633,
      "step": 3071
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.465791940689087,
      "learning_rate": 5.862916504609629e-06,
      "loss": 0.4877,
      "step": 3072
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5283629894256592,
      "learning_rate": 5.860520202457619e-06,
      "loss": 0.5212,
      "step": 3073
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1605925559997559,
      "learning_rate": 5.8581236966017604e-06,
      "loss": 0.4465,
      "step": 3074
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4302971363067627,
      "learning_rate": 5.855726987609359e-06,
      "loss": 0.5103,
      "step": 3075
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.258694052696228,
      "learning_rate": 5.853330076047764e-06,
      "loss": 0.4778,
      "step": 3076
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4321314096450806,
      "learning_rate": 5.850932962484381e-06,
      "loss": 0.5198,
      "step": 3077
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1800934076309204,
      "learning_rate": 5.848535647486659e-06,
      "loss": 0.7245,
      "step": 3078
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.027125597000122,
      "learning_rate": 5.846138131622091e-06,
      "loss": 0.5082,
      "step": 3079
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.072657823562622,
      "learning_rate": 5.843740415458223e-06,
      "loss": 0.375,
      "step": 3080
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3735058307647705,
      "learning_rate": 5.841342499562648e-06,
      "loss": 0.4242,
      "step": 3081
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.293311357498169,
      "learning_rate": 5.838944384503004e-06,
      "loss": 0.4171,
      "step": 3082
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4058811664581299,
      "learning_rate": 5.836546070846975e-06,
      "loss": 0.4669,
      "step": 3083
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0619202852249146,
      "learning_rate": 5.834147559162297e-06,
      "loss": 0.4789,
      "step": 3084
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.356904149055481,
      "learning_rate": 5.831748850016747e-06,
      "loss": 0.4805,
      "step": 3085
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5524216890335083,
      "learning_rate": 5.8293499439781535e-06,
      "loss": 0.4612,
      "step": 3086
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2388393878936768,
      "learning_rate": 5.8269508416143896e-06,
      "loss": 0.5051,
      "step": 3087
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0854469537734985,
      "learning_rate": 5.824551543493375e-06,
      "loss": 0.4429,
      "step": 3088
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.342893362045288,
      "learning_rate": 5.822152050183075e-06,
      "loss": 0.5318,
      "step": 3089
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5838475227355957,
      "learning_rate": 5.819752362251503e-06,
      "loss": 0.4852,
      "step": 3090
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0308094024658203,
      "learning_rate": 5.817352480266716e-06,
      "loss": 0.4363,
      "step": 3091
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.235683798789978,
      "learning_rate": 5.81495240479682e-06,
      "loss": 0.452,
      "step": 3092
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.928335964679718,
      "learning_rate": 5.812552136409962e-06,
      "loss": 0.3764,
      "step": 3093
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1420202255249023,
      "learning_rate": 5.8101516756743416e-06,
      "loss": 0.4583,
      "step": 3094
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0110571384429932,
      "learning_rate": 5.807751023158198e-06,
      "loss": 0.4948,
      "step": 3095
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0596953630447388,
      "learning_rate": 5.805350179429817e-06,
      "loss": 0.4197,
      "step": 3096
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1788299083709717,
      "learning_rate": 5.802949145057533e-06,
      "loss": 0.4269,
      "step": 3097
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7820701599121094,
      "learning_rate": 5.8005479206097205e-06,
      "loss": 0.4538,
      "step": 3098
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1585787534713745,
      "learning_rate": 5.798146506654803e-06,
      "loss": 0.5332,
      "step": 3099
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.016280174255371,
      "learning_rate": 5.795744903761248e-06,
      "loss": 0.486,
      "step": 3100
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.357146978378296,
      "learning_rate": 5.793343112497565e-06,
      "loss": 0.5148,
      "step": 3101
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5999542474746704,
      "learning_rate": 5.790941133432312e-06,
      "loss": 0.5094,
      "step": 3102
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4390569925308228,
      "learning_rate": 5.788538967134089e-06,
      "loss": 0.4688,
      "step": 3103
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1088740825653076,
      "learning_rate": 5.786136614171542e-06,
      "loss": 0.4323,
      "step": 3104
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.265368103981018,
      "learning_rate": 5.783734075113359e-06,
      "loss": 0.4636,
      "step": 3105
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2088619470596313,
      "learning_rate": 5.781331350528273e-06,
      "loss": 0.4754,
      "step": 3106
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3459906578063965,
      "learning_rate": 5.778928440985063e-06,
      "loss": 0.4648,
      "step": 3107
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1947335004806519,
      "learning_rate": 5.7765253470525485e-06,
      "loss": 0.382,
      "step": 3108
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1427867412567139,
      "learning_rate": 5.774122069299593e-06,
      "loss": 0.4668,
      "step": 3109
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2392992973327637,
      "learning_rate": 5.771718608295108e-06,
      "loss": 0.5671,
      "step": 3110
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6933894157409668,
      "learning_rate": 5.769314964608042e-06,
      "loss": 0.509,
      "step": 3111
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1927505731582642,
      "learning_rate": 5.76691113880739e-06,
      "loss": 0.4904,
      "step": 3112
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.714858055114746,
      "learning_rate": 5.764507131462192e-06,
      "loss": 0.4937,
      "step": 3113
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6507278680801392,
      "learning_rate": 5.762102943141526e-06,
      "loss": 0.4414,
      "step": 3114
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.382660984992981,
      "learning_rate": 5.759698574414519e-06,
      "loss": 0.411,
      "step": 3115
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4630095958709717,
      "learning_rate": 5.757294025850335e-06,
      "loss": 0.4767,
      "step": 3116
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2300593852996826,
      "learning_rate": 5.754889298018183e-06,
      "loss": 0.4831,
      "step": 3117
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1296401023864746,
      "learning_rate": 5.752484391487314e-06,
      "loss": 0.4779,
      "step": 3118
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4540261030197144,
      "learning_rate": 5.750079306827023e-06,
      "loss": 0.4906,
      "step": 3119
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2714569568634033,
      "learning_rate": 5.7476740446066445e-06,
      "loss": 0.459,
      "step": 3120
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4752650260925293,
      "learning_rate": 5.745268605395559e-06,
      "loss": 0.5377,
      "step": 3121
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7424825429916382,
      "learning_rate": 5.7428629897631835e-06,
      "loss": 0.3664,
      "step": 3122
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0747565031051636,
      "learning_rate": 5.7404571982789825e-06,
      "loss": 0.501,
      "step": 3123
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.5401248931884766,
      "learning_rate": 5.738051231512455e-06,
      "loss": 0.5746,
      "step": 3124
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3889168500900269,
      "learning_rate": 5.735645090033148e-06,
      "loss": 0.4893,
      "step": 3125
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1531027555465698,
      "learning_rate": 5.733238774410647e-06,
      "loss": 0.7522,
      "step": 3126
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3467766046524048,
      "learning_rate": 5.73083228521458e-06,
      "loss": 0.4496,
      "step": 3127
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2024993896484375,
      "learning_rate": 5.728425623014616e-06,
      "loss": 0.4453,
      "step": 3128
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4977807998657227,
      "learning_rate": 5.726018788380464e-06,
      "loss": 0.4283,
      "step": 3129
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2439889907836914,
      "learning_rate": 5.723611781881871e-06,
      "loss": 0.486,
      "step": 3130
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3818689584732056,
      "learning_rate": 5.721204604088631e-06,
      "loss": 0.5458,
      "step": 3131
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.140946388244629,
      "learning_rate": 5.718797255570575e-06,
      "loss": 0.3805,
      "step": 3132
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.140318751335144,
      "learning_rate": 5.716389736897574e-06,
      "loss": 0.4945,
      "step": 3133
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.117175817489624,
      "learning_rate": 5.71398204863954e-06,
      "loss": 0.4542,
      "step": 3134
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2297958135604858,
      "learning_rate": 5.711574191366427e-06,
      "loss": 0.4993,
      "step": 3135
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2797528505325317,
      "learning_rate": 5.7091661656482245e-06,
      "loss": 0.4749,
      "step": 3136
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.213875651359558,
      "learning_rate": 5.7067579720549645e-06,
      "loss": 0.5203,
      "step": 3137
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.326390027999878,
      "learning_rate": 5.704349611156722e-06,
      "loss": 0.5267,
      "step": 3138
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.151141881942749,
      "learning_rate": 5.701941083523606e-06,
      "loss": 0.7124,
      "step": 3139
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4342890977859497,
      "learning_rate": 5.699532389725767e-06,
      "loss": 0.5473,
      "step": 3140
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1959879398345947,
      "learning_rate": 5.697123530333397e-06,
      "loss": 0.4632,
      "step": 3141
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2901360988616943,
      "learning_rate": 5.694714505916723e-06,
      "loss": 0.4989,
      "step": 3142
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4628450870513916,
      "learning_rate": 5.692305317046016e-06,
      "loss": 0.5327,
      "step": 3143
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3105391263961792,
      "learning_rate": 5.689895964291583e-06,
      "loss": 0.5111,
      "step": 3144
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4367220401763916,
      "learning_rate": 5.687486448223769e-06,
      "loss": 0.515,
      "step": 3145
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1934125423431396,
      "learning_rate": 5.68507676941296e-06,
      "loss": 0.4965,
      "step": 3146
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4034509658813477,
      "learning_rate": 5.682666928429579e-06,
      "loss": 0.5254,
      "step": 3147
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3398991823196411,
      "learning_rate": 5.680256925844085e-06,
      "loss": 0.4888,
      "step": 3148
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.365146517753601,
      "learning_rate": 5.677846762226982e-06,
      "loss": 0.5032,
      "step": 3149
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.535241961479187,
      "learning_rate": 5.675436438148807e-06,
      "loss": 0.499,
      "step": 3150
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.2383289337158203,
      "learning_rate": 5.673025954180136e-06,
      "loss": 0.5782,
      "step": 3151
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0863618850708008,
      "learning_rate": 5.670615310891583e-06,
      "loss": 0.7066,
      "step": 3152
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2020039558410645,
      "learning_rate": 5.668204508853798e-06,
      "loss": 0.4948,
      "step": 3153
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1867098808288574,
      "learning_rate": 5.6657935486374735e-06,
      "loss": 0.4591,
      "step": 3154
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6328516006469727,
      "learning_rate": 5.6633824308133344e-06,
      "loss": 0.5251,
      "step": 3155
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4031263589859009,
      "learning_rate": 5.660971155952145e-06,
      "loss": 0.4818,
      "step": 3156
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0355316400527954,
      "learning_rate": 5.658559724624706e-06,
      "loss": 0.5138,
      "step": 3157
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4239869117736816,
      "learning_rate": 5.6561481374018565e-06,
      "loss": 0.4481,
      "step": 3158
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.233300805091858,
      "learning_rate": 5.653736394854471e-06,
      "loss": 0.4392,
      "step": 3159
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6441513299942017,
      "learning_rate": 5.651324497553462e-06,
      "loss": 0.4695,
      "step": 3160
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0400781631469727,
      "learning_rate": 5.648912446069777e-06,
      "loss": 0.4406,
      "step": 3161
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0369631052017212,
      "learning_rate": 5.646500240974402e-06,
      "loss": 0.4439,
      "step": 3162
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3207242488861084,
      "learning_rate": 5.644087882838358e-06,
      "loss": 0.3936,
      "step": 3163
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5881688594818115,
      "learning_rate": 5.6416753722327025e-06,
      "loss": 0.4211,
      "step": 3164
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.138211965560913,
      "learning_rate": 5.63926270972853e-06,
      "loss": 0.4688,
      "step": 3165
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1874315738677979,
      "learning_rate": 5.63684989589697e-06,
      "loss": 0.4785,
      "step": 3166
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6061674356460571,
      "learning_rate": 5.6344369313091865e-06,
      "loss": 0.4403,
      "step": 3167
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2668076753616333,
      "learning_rate": 5.632023816536382e-06,
      "loss": 0.3789,
      "step": 3168
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2231875658035278,
      "learning_rate": 5.629610552149795e-06,
      "loss": 0.5068,
      "step": 3169
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1265228986740112,
      "learning_rate": 5.627197138720695e-06,
      "loss": 0.4507,
      "step": 3170
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.447356939315796,
      "learning_rate": 5.624783576820389e-06,
      "loss": 0.4972,
      "step": 3171
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.710938811302185,
      "learning_rate": 5.622369867020222e-06,
      "loss": 0.4283,
      "step": 3172
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.420427918434143,
      "learning_rate": 5.61995600989157e-06,
      "loss": 0.4582,
      "step": 3173
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3134490251541138,
      "learning_rate": 5.617542006005847e-06,
      "loss": 0.4449,
      "step": 3174
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.383414626121521,
      "learning_rate": 5.615127855934499e-06,
      "loss": 0.4807,
      "step": 3175
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0680228471755981,
      "learning_rate": 5.612713560249008e-06,
      "loss": 0.4771,
      "step": 3176
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4743345975875854,
      "learning_rate": 5.61029911952089e-06,
      "loss": 0.4526,
      "step": 3177
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.844527006149292,
      "learning_rate": 5.607884534321696e-06,
      "loss": 0.475,
      "step": 3178
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0665446519851685,
      "learning_rate": 5.605469805223011e-06,
      "loss": 0.7098,
      "step": 3179
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.286661982536316,
      "learning_rate": 5.603054932796453e-06,
      "loss": 0.4595,
      "step": 3180
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6884136199951172,
      "learning_rate": 5.600639917613675e-06,
      "loss": 0.5381,
      "step": 3181
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2122164964675903,
      "learning_rate": 5.5982247602463655e-06,
      "loss": 0.737,
      "step": 3182
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5426965951919556,
      "learning_rate": 5.595809461266244e-06,
      "loss": 0.4645,
      "step": 3183
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.440367579460144,
      "learning_rate": 5.5933940212450635e-06,
      "loss": 0.4974,
      "step": 3184
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.664681911468506,
      "learning_rate": 5.590978440754611e-06,
      "loss": 0.4908,
      "step": 3185
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7093584537506104,
      "learning_rate": 5.588562720366707e-06,
      "loss": 0.488,
      "step": 3186
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.469010591506958,
      "learning_rate": 5.586146860653205e-06,
      "loss": 0.5027,
      "step": 3187
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3453456163406372,
      "learning_rate": 5.583730862185993e-06,
      "loss": 0.4978,
      "step": 3188
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1214373111724854,
      "learning_rate": 5.5813147255369894e-06,
      "loss": 0.4406,
      "step": 3189
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8169537782669067,
      "learning_rate": 5.578898451278145e-06,
      "loss": 0.4569,
      "step": 3190
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1353098154067993,
      "learning_rate": 5.576482039981448e-06,
      "loss": 0.4364,
      "step": 3191
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5369378328323364,
      "learning_rate": 5.574065492218911e-06,
      "loss": 0.477,
      "step": 3192
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8096050024032593,
      "learning_rate": 5.571648808562587e-06,
      "loss": 0.5624,
      "step": 3193
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.262475609779358,
      "learning_rate": 5.569231989584555e-06,
      "loss": 0.4378,
      "step": 3194
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3677845001220703,
      "learning_rate": 5.566815035856932e-06,
      "loss": 0.4496,
      "step": 3195
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2458970546722412,
      "learning_rate": 5.564397947951859e-06,
      "loss": 0.567,
      "step": 3196
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1560784578323364,
      "learning_rate": 5.5619807264415185e-06,
      "loss": 0.4164,
      "step": 3197
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.10827898979187,
      "learning_rate": 5.559563371898114e-06,
      "loss": 0.4892,
      "step": 3198
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3054678440093994,
      "learning_rate": 5.557145884893889e-06,
      "loss": 0.4439,
      "step": 3199
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4563730955123901,
      "learning_rate": 5.554728266001115e-06,
      "loss": 0.5395,
      "step": 3200
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3375327587127686,
      "learning_rate": 5.5523105157920956e-06,
      "loss": 0.5078,
      "step": 3201
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4922235012054443,
      "learning_rate": 5.549892634839163e-06,
      "loss": 0.3996,
      "step": 3202
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4406849145889282,
      "learning_rate": 5.547474623714684e-06,
      "loss": 0.4754,
      "step": 3203
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3850826025009155,
      "learning_rate": 5.545056482991051e-06,
      "loss": 0.4947,
      "step": 3204
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.340912103652954,
      "learning_rate": 5.542638213240697e-06,
      "loss": 0.7544,
      "step": 3205
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1452383995056152,
      "learning_rate": 5.540219815036072e-06,
      "loss": 0.4082,
      "step": 3206
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2268128395080566,
      "learning_rate": 5.537801288949669e-06,
      "loss": 0.5454,
      "step": 3207
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.285654902458191,
      "learning_rate": 5.535382635554003e-06,
      "loss": 0.4795,
      "step": 3208
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5907710790634155,
      "learning_rate": 5.53296385542162e-06,
      "loss": 0.5046,
      "step": 3209
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8914341926574707,
      "learning_rate": 5.530544949125102e-06,
      "loss": 0.4864,
      "step": 3210
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3885937929153442,
      "learning_rate": 5.5281259172370536e-06,
      "loss": 0.4887,
      "step": 3211
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4071173667907715,
      "learning_rate": 5.525706760330115e-06,
      "loss": 0.4863,
      "step": 3212
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2718101739883423,
      "learning_rate": 5.523287478976952e-06,
      "loss": 0.5471,
      "step": 3213
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3868544101715088,
      "learning_rate": 5.520868073750261e-06,
      "loss": 0.5026,
      "step": 3214
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1756550073623657,
      "learning_rate": 5.518448545222765e-06,
      "loss": 0.4993,
      "step": 3215
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1049330234527588,
      "learning_rate": 5.516028893967223e-06,
      "loss": 0.4791,
      "step": 3216
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3702163696289062,
      "learning_rate": 5.513609120556418e-06,
      "loss": 0.5136,
      "step": 3217
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1100060939788818,
      "learning_rate": 5.5111892255631635e-06,
      "loss": 0.5206,
      "step": 3218
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.507979154586792,
      "learning_rate": 5.508769209560302e-06,
      "loss": 0.4797,
      "step": 3219
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.191996455192566,
      "learning_rate": 5.506349073120699e-06,
      "loss": 0.4289,
      "step": 3220
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4521008729934692,
      "learning_rate": 5.503928816817259e-06,
      "loss": 0.4923,
      "step": 3221
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.231505274772644,
      "learning_rate": 5.501508441222906e-06,
      "loss": 0.5548,
      "step": 3222
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3176743984222412,
      "learning_rate": 5.499087946910598e-06,
      "loss": 0.4987,
      "step": 3223
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.107932686805725,
      "learning_rate": 5.496667334453317e-06,
      "loss": 0.4733,
      "step": 3224
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.037619709968567,
      "learning_rate": 5.494246604424076e-06,
      "loss": 0.4349,
      "step": 3225
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1661655902862549,
      "learning_rate": 5.491825757395912e-06,
      "loss": 0.3819,
      "step": 3226
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3819814920425415,
      "learning_rate": 5.489404793941895e-06,
      "loss": 0.5007,
      "step": 3227
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.981391191482544,
      "learning_rate": 5.486983714635118e-06,
      "loss": 0.4742,
      "step": 3228
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1629717350006104,
      "learning_rate": 5.484562520048703e-06,
      "loss": 0.449,
      "step": 3229
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0755656957626343,
      "learning_rate": 5.4821412107558005e-06,
      "loss": 0.7224,
      "step": 3230
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.233819842338562,
      "learning_rate": 5.479719787329586e-06,
      "loss": 0.4743,
      "step": 3231
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5407261848449707,
      "learning_rate": 5.477298250343264e-06,
      "loss": 0.5081,
      "step": 3232
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.14481520652771,
      "learning_rate": 5.474876600370064e-06,
      "loss": 0.5627,
      "step": 3233
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5059961080551147,
      "learning_rate": 5.472454837983244e-06,
      "loss": 0.453,
      "step": 3234
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4403494596481323,
      "learning_rate": 5.470032963756087e-06,
      "loss": 0.4584,
      "step": 3235
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.317753553390503,
      "learning_rate": 5.467610978261906e-06,
      "loss": 0.5183,
      "step": 3236
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.520492672920227,
      "learning_rate": 5.465188882074035e-06,
      "loss": 0.4958,
      "step": 3237
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5375938415527344,
      "learning_rate": 5.462766675765836e-06,
      "loss": 0.7667,
      "step": 3238
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0983867645263672,
      "learning_rate": 5.460344359910701e-06,
      "loss": 0.7364,
      "step": 3239
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.979192316532135,
      "learning_rate": 5.457921935082042e-06,
      "loss": 0.4956,
      "step": 3240
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.322526454925537,
      "learning_rate": 5.455499401853301e-06,
      "loss": 0.5055,
      "step": 3241
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.062259554862976,
      "learning_rate": 5.453076760797945e-06,
      "loss": 0.6794,
      "step": 3242
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1857354640960693,
      "learning_rate": 5.4506540124894646e-06,
      "loss": 0.505,
      "step": 3243
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3603730201721191,
      "learning_rate": 5.448231157501377e-06,
      "loss": 0.4635,
      "step": 3244
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.029020309448242,
      "learning_rate": 5.445808196407225e-06,
      "loss": 0.4014,
      "step": 3245
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0806481838226318,
      "learning_rate": 5.4433851297805764e-06,
      "loss": 0.5569,
      "step": 3246
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1475685834884644,
      "learning_rate": 5.440961958195025e-06,
      "loss": 0.4833,
      "step": 3247
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.275903344154358,
      "learning_rate": 5.438538682224185e-06,
      "loss": 0.5578,
      "step": 3248
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.141414761543274,
      "learning_rate": 5.436115302441702e-06,
      "loss": 0.5042,
      "step": 3249
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.235708236694336,
      "learning_rate": 5.43369181942124e-06,
      "loss": 0.5144,
      "step": 3250
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9903731942176819,
      "learning_rate": 5.431268233736491e-06,
      "loss": 0.4269,
      "step": 3251
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.503527045249939,
      "learning_rate": 5.42884454596117e-06,
      "loss": 0.4197,
      "step": 3252
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.188199758529663,
      "learning_rate": 5.426420756669019e-06,
      "loss": 0.4546,
      "step": 3253
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2036619186401367,
      "learning_rate": 5.4239968664338e-06,
      "loss": 0.4784,
      "step": 3254
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2389755249023438,
      "learning_rate": 5.421572875829299e-06,
      "loss": 0.4715,
      "step": 3255
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.12288498878479,
      "learning_rate": 5.41914878542933e-06,
      "loss": 0.4216,
      "step": 3256
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0608559846878052,
      "learning_rate": 5.416724595807726e-06,
      "loss": 0.4803,
      "step": 3257
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4425976276397705,
      "learning_rate": 5.414300307538347e-06,
      "loss": 0.4477,
      "step": 3258
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.190598726272583,
      "learning_rate": 5.4118759211950726e-06,
      "loss": 0.413,
      "step": 3259
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.376530647277832,
      "learning_rate": 5.4094514373518104e-06,
      "loss": 0.4841,
      "step": 3260
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.598429799079895,
      "learning_rate": 5.407026856582487e-06,
      "loss": 0.541,
      "step": 3261
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6868913173675537,
      "learning_rate": 5.404602179461054e-06,
      "loss": 0.5186,
      "step": 3262
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.007330298423767,
      "learning_rate": 5.402177406561483e-06,
      "loss": 0.4179,
      "step": 3263
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4326200485229492,
      "learning_rate": 5.399752538457773e-06,
      "loss": 0.4848,
      "step": 3264
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9625905752182007,
      "learning_rate": 5.397327575723942e-06,
      "loss": 0.3744,
      "step": 3265
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2005279064178467,
      "learning_rate": 5.394902518934031e-06,
      "loss": 0.4159,
      "step": 3266
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.478883147239685,
      "learning_rate": 5.392477368662107e-06,
      "loss": 0.4911,
      "step": 3267
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1505818367004395,
      "learning_rate": 5.390052125482252e-06,
      "loss": 0.3777,
      "step": 3268
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.336296796798706,
      "learning_rate": 5.387626789968574e-06,
      "loss": 0.4993,
      "step": 3269
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.9741251468658447,
      "learning_rate": 5.385201362695204e-06,
      "loss": 0.5234,
      "step": 3270
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.423726201057434,
      "learning_rate": 5.382775844236293e-06,
      "loss": 0.5281,
      "step": 3271
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1832681894302368,
      "learning_rate": 5.380350235166014e-06,
      "loss": 0.4941,
      "step": 3272
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.433073878288269,
      "learning_rate": 5.377924536058563e-06,
      "loss": 0.4806,
      "step": 3273
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.440516710281372,
      "learning_rate": 5.375498747488153e-06,
      "loss": 0.5092,
      "step": 3274
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2669707536697388,
      "learning_rate": 5.373072870029022e-06,
      "loss": 0.4061,
      "step": 3275
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.406273603439331,
      "learning_rate": 5.370646904255428e-06,
      "loss": 0.4788,
      "step": 3276
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2789924144744873,
      "learning_rate": 5.36822085074165e-06,
      "loss": 0.4757,
      "step": 3277
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3715063333511353,
      "learning_rate": 5.365794710061987e-06,
      "loss": 0.481,
      "step": 3278
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5391491651535034,
      "learning_rate": 5.363368482790762e-06,
      "loss": 0.5045,
      "step": 3279
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8849588632583618,
      "learning_rate": 5.360942169502312e-06,
      "loss": 0.3117,
      "step": 3280
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.1273090839385986,
      "learning_rate": 5.358515770770999e-06,
      "loss": 0.488,
      "step": 3281
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4407739639282227,
      "learning_rate": 5.356089287171206e-06,
      "loss": 0.4833,
      "step": 3282
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.296326756477356,
      "learning_rate": 5.353662719277334e-06,
      "loss": 0.4718,
      "step": 3283
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1765350103378296,
      "learning_rate": 5.351236067663803e-06,
      "loss": 0.5121,
      "step": 3284
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1423487663269043,
      "learning_rate": 5.348809332905056e-06,
      "loss": 0.4839,
      "step": 3285
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2322319746017456,
      "learning_rate": 5.346382515575553e-06,
      "loss": 0.4164,
      "step": 3286
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2797902822494507,
      "learning_rate": 5.343955616249773e-06,
      "loss": 0.5414,
      "step": 3287
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.356561303138733,
      "learning_rate": 5.341528635502219e-06,
      "loss": 0.4513,
      "step": 3288
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.245186686515808,
      "learning_rate": 5.3391015739074085e-06,
      "loss": 0.4579,
      "step": 3289
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3131670951843262,
      "learning_rate": 5.336674432039879e-06,
      "loss": 0.4104,
      "step": 3290
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5377042293548584,
      "learning_rate": 5.334247210474189e-06,
      "loss": 0.4677,
      "step": 3291
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5130728483200073,
      "learning_rate": 5.3318199097849146e-06,
      "loss": 0.5233,
      "step": 3292
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2750556468963623,
      "learning_rate": 5.329392530546649e-06,
      "loss": 0.4652,
      "step": 3293
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.996993064880371,
      "learning_rate": 5.326965073334007e-06,
      "loss": 0.4944,
      "step": 3294
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1659142971038818,
      "learning_rate": 5.324537538721621e-06,
      "loss": 0.4931,
      "step": 3295
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1831905841827393,
      "learning_rate": 5.322109927284142e-06,
      "loss": 0.4781,
      "step": 3296
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9283664226531982,
      "learning_rate": 5.3196822395962355e-06,
      "loss": 0.4543,
      "step": 3297
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4451323747634888,
      "learning_rate": 5.31725447623259e-06,
      "loss": 0.4979,
      "step": 3298
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0297983884811401,
      "learning_rate": 5.314826637767909e-06,
      "loss": 0.4619,
      "step": 3299
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9435991048812866,
      "learning_rate": 5.312398724776915e-06,
      "loss": 0.447,
      "step": 3300
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.178680658340454,
      "learning_rate": 5.309970737834349e-06,
      "loss": 0.4715,
      "step": 3301
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3105436563491821,
      "learning_rate": 5.3075426775149665e-06,
      "loss": 0.427,
      "step": 3302
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1918885707855225,
      "learning_rate": 5.305114544393544e-06,
      "loss": 0.4584,
      "step": 3303
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3938292264938354,
      "learning_rate": 5.3026863390448705e-06,
      "loss": 0.3976,
      "step": 3304
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4819977283477783,
      "learning_rate": 5.300258062043758e-06,
      "loss": 0.3968,
      "step": 3305
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4352387189865112,
      "learning_rate": 5.2978297139650314e-06,
      "loss": 0.5122,
      "step": 3306
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9966845512390137,
      "learning_rate": 5.2954012953835334e-06,
      "loss": 0.5201,
      "step": 3307
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2009069919586182,
      "learning_rate": 5.292972806874123e-06,
      "loss": 0.4517,
      "step": 3308
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2011905908584595,
      "learning_rate": 5.290544249011678e-06,
      "loss": 0.5045,
      "step": 3309
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3466640710830688,
      "learning_rate": 5.28811562237109e-06,
      "loss": 0.4993,
      "step": 3310
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1494559049606323,
      "learning_rate": 5.285686927527266e-06,
      "loss": 0.422,
      "step": 3311
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2936896085739136,
      "learning_rate": 5.283258165055134e-06,
      "loss": 0.4874,
      "step": 3312
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3061695098876953,
      "learning_rate": 5.280829335529632e-06,
      "loss": 0.4644,
      "step": 3313
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3326125144958496,
      "learning_rate": 5.278400439525718e-06,
      "loss": 0.494,
      "step": 3314
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.194596529006958,
      "learning_rate": 5.275971477618366e-06,
      "loss": 0.4279,
      "step": 3315
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1805810928344727,
      "learning_rate": 5.273542450382561e-06,
      "loss": 0.4886,
      "step": 3316
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4561398029327393,
      "learning_rate": 5.271113358393309e-06,
      "loss": 0.5008,
      "step": 3317
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2037140130996704,
      "learning_rate": 5.268684202225628e-06,
      "loss": 0.3906,
      "step": 3318
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4184749126434326,
      "learning_rate": 5.266254982454553e-06,
      "loss": 0.4712,
      "step": 3319
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4890929460525513,
      "learning_rate": 5.2638256996551305e-06,
      "loss": 0.4921,
      "step": 3320
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0319018363952637,
      "learning_rate": 5.261396354402428e-06,
      "loss": 0.4221,
      "step": 3321
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1787306070327759,
      "learning_rate": 5.258966947271524e-06,
      "loss": 0.5001,
      "step": 3322
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1807432174682617,
      "learning_rate": 5.256537478837509e-06,
      "loss": 0.4722,
      "step": 3323
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.298539638519287,
      "learning_rate": 5.254107949675493e-06,
      "loss": 0.57,
      "step": 3324
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8089324235916138,
      "learning_rate": 5.251678360360599e-06,
      "loss": 0.4454,
      "step": 3325
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5970649719238281,
      "learning_rate": 5.249248711467963e-06,
      "loss": 0.4858,
      "step": 3326
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.334078073501587,
      "learning_rate": 5.246819003572735e-06,
      "loss": 0.5116,
      "step": 3327
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3631625175476074,
      "learning_rate": 5.2443892372500805e-06,
      "loss": 0.5047,
      "step": 3328
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4679384231567383,
      "learning_rate": 5.241959413075178e-06,
      "loss": 0.5173,
      "step": 3329
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1738972663879395,
      "learning_rate": 5.23952953162322e-06,
      "loss": 0.5,
      "step": 3330
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.293351173400879,
      "learning_rate": 5.237099593469411e-06,
      "loss": 0.429,
      "step": 3331
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.150368094444275,
      "learning_rate": 5.234669599188972e-06,
      "loss": 0.454,
      "step": 3332
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4569852352142334,
      "learning_rate": 5.232239549357134e-06,
      "loss": 0.5095,
      "step": 3333
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1992380619049072,
      "learning_rate": 5.2298094445491444e-06,
      "loss": 0.4685,
      "step": 3334
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.300737738609314,
      "learning_rate": 5.227379285340259e-06,
      "loss": 0.4702,
      "step": 3335
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.107317328453064,
      "learning_rate": 5.2249490723057525e-06,
      "loss": 0.4696,
      "step": 3336
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9015408754348755,
      "learning_rate": 5.222518806020907e-06,
      "loss": 0.5528,
      "step": 3337
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0223959684371948,
      "learning_rate": 5.22008848706102e-06,
      "loss": 0.3814,
      "step": 3338
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.017937183380127,
      "learning_rate": 5.217658116001404e-06,
      "loss": 0.4091,
      "step": 3339
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.352066993713379,
      "learning_rate": 5.2152276934173755e-06,
      "loss": 0.5156,
      "step": 3340
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1968884468078613,
      "learning_rate": 5.21279721988427e-06,
      "loss": 0.5416,
      "step": 3341
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1362820863723755,
      "learning_rate": 5.210366695977435e-06,
      "loss": 0.4907,
      "step": 3342
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5142004489898682,
      "learning_rate": 5.2079361222722255e-06,
      "loss": 0.4583,
      "step": 3343
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4589754343032837,
      "learning_rate": 5.205505499344015e-06,
      "loss": 0.3988,
      "step": 3344
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1374026536941528,
      "learning_rate": 5.203074827768183e-06,
      "loss": 0.7123,
      "step": 3345
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.146177053451538,
      "learning_rate": 5.200644108120122e-06,
      "loss": 0.4735,
      "step": 3346
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1225769519805908,
      "learning_rate": 5.198213340975237e-06,
      "loss": 0.3959,
      "step": 3347
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.235103726387024,
      "learning_rate": 5.19578252690894e-06,
      "loss": 0.4497,
      "step": 3348
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.178352952003479,
      "learning_rate": 5.193351666496664e-06,
      "loss": 0.4296,
      "step": 3349
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3338712453842163,
      "learning_rate": 5.1909207603138414e-06,
      "loss": 0.4535,
      "step": 3350
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4319921731948853,
      "learning_rate": 5.188489808935924e-06,
      "loss": 0.4651,
      "step": 3351
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0661855936050415,
      "learning_rate": 5.186058812938367e-06,
      "loss": 0.4185,
      "step": 3352
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2895466089248657,
      "learning_rate": 5.183627772896642e-06,
      "loss": 0.7481,
      "step": 3353
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0513108968734741,
      "learning_rate": 5.181196689386229e-06,
      "loss": 0.4592,
      "step": 3354
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1990457773208618,
      "learning_rate": 5.17876556298262e-06,
      "loss": 0.6341,
      "step": 3355
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2686004638671875,
      "learning_rate": 5.176334394261312e-06,
      "loss": 0.4417,
      "step": 3356
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1755720376968384,
      "learning_rate": 5.17390318379782e-06,
      "loss": 0.551,
      "step": 3357
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3185782432556152,
      "learning_rate": 5.171471932167661e-06,
      "loss": 0.4784,
      "step": 3358
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.41685950756073,
      "learning_rate": 5.169040639946364e-06,
      "loss": 0.4729,
      "step": 3359
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.174736499786377,
      "learning_rate": 5.166609307709473e-06,
      "loss": 0.4527,
      "step": 3360
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2389800548553467,
      "learning_rate": 5.164177936032534e-06,
      "loss": 0.4908,
      "step": 3361
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0494859218597412,
      "learning_rate": 5.161746525491107e-06,
      "loss": 0.4347,
      "step": 3362
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4040398597717285,
      "learning_rate": 5.15931507666076e-06,
      "loss": 0.4803,
      "step": 3363
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9641011953353882,
      "learning_rate": 5.156883590117068e-06,
      "loss": 0.4622,
      "step": 3364
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2316597700119019,
      "learning_rate": 5.154452066435617e-06,
      "loss": 0.4211,
      "step": 3365
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2400368452072144,
      "learning_rate": 5.152020506192003e-06,
      "loss": 0.5156,
      "step": 3366
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2189667224884033,
      "learning_rate": 5.14958890996183e-06,
      "loss": 0.5135,
      "step": 3367
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2948092222213745,
      "learning_rate": 5.147157278320708e-06,
      "loss": 0.4685,
      "step": 3368
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9960840344429016,
      "learning_rate": 5.144725611844258e-06,
      "loss": 0.4577,
      "step": 3369
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0662883520126343,
      "learning_rate": 5.142293911108107e-06,
      "loss": 0.4987,
      "step": 3370
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6045466661453247,
      "learning_rate": 5.139862176687891e-06,
      "loss": 0.4588,
      "step": 3371
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3645098209381104,
      "learning_rate": 5.137430409159256e-06,
      "loss": 0.5523,
      "step": 3372
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2902960777282715,
      "learning_rate": 5.1349986090978555e-06,
      "loss": 0.4976,
      "step": 3373
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2960669994354248,
      "learning_rate": 5.132566777079347e-06,
      "loss": 0.3916,
      "step": 3374
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.550581693649292,
      "learning_rate": 5.130134913679401e-06,
      "loss": 0.4149,
      "step": 3375
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1421363353729248,
      "learning_rate": 5.127703019473686e-06,
      "loss": 0.405,
      "step": 3376
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2919256687164307,
      "learning_rate": 5.125271095037893e-06,
      "loss": 0.5156,
      "step": 3377
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.362741231918335,
      "learning_rate": 5.122839140947704e-06,
      "loss": 0.4682,
      "step": 3378
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3439819812774658,
      "learning_rate": 5.12040715777882e-06,
      "loss": 0.4869,
      "step": 3379
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4458913803100586,
      "learning_rate": 5.117975146106942e-06,
      "loss": 0.5084,
      "step": 3380
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.00521981716156,
      "learning_rate": 5.115543106507782e-06,
      "loss": 0.5444,
      "step": 3381
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0424288511276245,
      "learning_rate": 5.113111039557055e-06,
      "loss": 0.517,
      "step": 3382
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1479859352111816,
      "learning_rate": 5.110678945830485e-06,
      "loss": 0.4687,
      "step": 3383
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.17115318775177,
      "learning_rate": 5.108246825903801e-06,
      "loss": 0.4414,
      "step": 3384
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1581836938858032,
      "learning_rate": 5.105814680352739e-06,
      "loss": 0.4568,
      "step": 3385
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3706477880477905,
      "learning_rate": 5.1033825097530395e-06,
      "loss": 0.4831,
      "step": 3386
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0446597337722778,
      "learning_rate": 5.100950314680453e-06,
      "loss": 0.5097,
      "step": 3387
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.249995231628418,
      "learning_rate": 5.098518095710731e-06,
      "loss": 0.4851,
      "step": 3388
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.068913459777832,
      "learning_rate": 5.096085853419631e-06,
      "loss": 0.4286,
      "step": 3389
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1314828395843506,
      "learning_rate": 5.093653588382922e-06,
      "loss": 0.4072,
      "step": 3390
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.239456295967102,
      "learning_rate": 5.09122130117637e-06,
      "loss": 0.5,
      "step": 3391
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.142535924911499,
      "learning_rate": 5.0887889923757525e-06,
      "loss": 0.4635,
      "step": 3392
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.35124933719635,
      "learning_rate": 5.0863566625568495e-06,
      "loss": 0.519,
      "step": 3393
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0328946113586426,
      "learning_rate": 5.083924312295445e-06,
      "loss": 0.4393,
      "step": 3394
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4016149044036865,
      "learning_rate": 5.0814919421673315e-06,
      "loss": 0.5373,
      "step": 3395
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5713074207305908,
      "learning_rate": 5.079059552748302e-06,
      "loss": 0.5234,
      "step": 3396
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0334678888320923,
      "learning_rate": 5.076627144614158e-06,
      "loss": 0.4708,
      "step": 3397
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.091280221939087,
      "learning_rate": 5.074194718340701e-06,
      "loss": 0.4788,
      "step": 3398
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4043171405792236,
      "learning_rate": 5.071762274503741e-06,
      "loss": 0.4965,
      "step": 3399
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9783814549446106,
      "learning_rate": 5.0693298136790895e-06,
      "loss": 0.4683,
      "step": 3400
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7862951755523682,
      "learning_rate": 5.066897336442563e-06,
      "loss": 0.4638,
      "step": 3401
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1486153602600098,
      "learning_rate": 5.064464843369983e-06,
      "loss": 0.4468,
      "step": 3402
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2477189302444458,
      "learning_rate": 5.062032335037171e-06,
      "loss": 0.4333,
      "step": 3403
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0941659212112427,
      "learning_rate": 5.0595998120199565e-06,
      "loss": 0.4627,
      "step": 3404
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0241144895553589,
      "learning_rate": 5.0571672748941705e-06,
      "loss": 0.3955,
      "step": 3405
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8799197673797607,
      "learning_rate": 5.054734724235647e-06,
      "loss": 0.4845,
      "step": 3406
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1474144458770752,
      "learning_rate": 5.052302160620223e-06,
      "loss": 0.4647,
      "step": 3407
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.249284029006958,
      "learning_rate": 5.049869584623741e-06,
      "loss": 0.4752,
      "step": 3408
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2959498167037964,
      "learning_rate": 5.047436996822043e-06,
      "loss": 0.4611,
      "step": 3409
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1836491823196411,
      "learning_rate": 5.045004397790974e-06,
      "loss": 0.4908,
      "step": 3410
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3847960233688354,
      "learning_rate": 5.042571788106389e-06,
      "loss": 0.4569,
      "step": 3411
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.386151909828186,
      "learning_rate": 5.040139168344135e-06,
      "loss": 0.4846,
      "step": 3412
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1741979122161865,
      "learning_rate": 5.037706539080066e-06,
      "loss": 0.4717,
      "step": 3413
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1548634767532349,
      "learning_rate": 5.035273900890039e-06,
      "loss": 0.4537,
      "step": 3414
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3115609884262085,
      "learning_rate": 5.032841254349911e-06,
      "loss": 0.4594,
      "step": 3415
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2408993244171143,
      "learning_rate": 5.030408600035546e-06,
      "loss": 0.4834,
      "step": 3416
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2273436784744263,
      "learning_rate": 5.0279759385228045e-06,
      "loss": 0.5079,
      "step": 3417
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0341451168060303,
      "learning_rate": 5.025543270387548e-06,
      "loss": 0.4671,
      "step": 3418
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2918399572372437,
      "learning_rate": 5.0231105962056445e-06,
      "loss": 0.5187,
      "step": 3419
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3443450927734375,
      "learning_rate": 5.020677916552959e-06,
      "loss": 0.5267,
      "step": 3420
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2373913526535034,
      "learning_rate": 5.018245232005362e-06,
      "loss": 0.4305,
      "step": 3421
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.256072998046875,
      "learning_rate": 5.015812543138722e-06,
      "loss": 0.4607,
      "step": 3422
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0820714235305786,
      "learning_rate": 5.01337985052891e-06,
      "loss": 0.5034,
      "step": 3423
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1881935596466064,
      "learning_rate": 5.0109471547517954e-06,
      "loss": 0.4543,
      "step": 3424
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.361560583114624,
      "learning_rate": 5.00851445638325e-06,
      "loss": 0.5174,
      "step": 3425
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1179683208465576,
      "learning_rate": 5.0060817559991484e-06,
      "loss": 0.4767,
      "step": 3426
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8780657052993774,
      "learning_rate": 5.003649054175362e-06,
      "loss": 0.4734,
      "step": 3427
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0128408670425415,
      "learning_rate": 5.001216351487767e-06,
      "loss": 0.4696,
      "step": 3428
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3463635444641113,
      "learning_rate": 4.998783648512235e-06,
      "loss": 0.5188,
      "step": 3429
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1976046562194824,
      "learning_rate": 4.996350945824639e-06,
      "loss": 0.7164,
      "step": 3430
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2887734174728394,
      "learning_rate": 4.993918244000854e-06,
      "loss": 0.5053,
      "step": 3431
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2823489904403687,
      "learning_rate": 4.991485543616752e-06,
      "loss": 0.4806,
      "step": 3432
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.035496473312378,
      "learning_rate": 4.989052845248208e-06,
      "loss": 0.4896,
      "step": 3433
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2113962173461914,
      "learning_rate": 4.9866201494710934e-06,
      "loss": 0.4451,
      "step": 3434
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2109615802764893,
      "learning_rate": 4.984187456861278e-06,
      "loss": 0.4738,
      "step": 3435
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.180864691734314,
      "learning_rate": 4.981754767994638e-06,
      "loss": 0.4216,
      "step": 3436
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.115186095237732,
      "learning_rate": 4.979322083447042e-06,
      "loss": 0.7049,
      "step": 3437
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.221832036972046,
      "learning_rate": 4.976889403794357e-06,
      "loss": 0.5248,
      "step": 3438
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9607317447662354,
      "learning_rate": 4.9744567296124536e-06,
      "loss": 0.4468,
      "step": 3439
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3902608156204224,
      "learning_rate": 4.972024061477197e-06,
      "loss": 0.4719,
      "step": 3440
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0448716878890991,
      "learning_rate": 4.969591399964456e-06,
      "loss": 0.3849,
      "step": 3441
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0589765310287476,
      "learning_rate": 4.96715874565009e-06,
      "loss": 0.4433,
      "step": 3442
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0464210510253906,
      "learning_rate": 4.9647260991099635e-06,
      "loss": 0.3448,
      "step": 3443
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0683612823486328,
      "learning_rate": 4.9622934609199365e-06,
      "loss": 0.5055,
      "step": 3444
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3587212562561035,
      "learning_rate": 4.959860831655867e-06,
      "loss": 0.4137,
      "step": 3445
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.115384817123413,
      "learning_rate": 4.957428211893611e-06,
      "loss": 0.4162,
      "step": 3446
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0735316276550293,
      "learning_rate": 4.954995602209025e-06,
      "loss": 0.4665,
      "step": 3447
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0123100280761719,
      "learning_rate": 4.952563003177959e-06,
      "loss": 0.4718,
      "step": 3448
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.677572250366211,
      "learning_rate": 4.950130415376261e-06,
      "loss": 0.4228,
      "step": 3449
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1003453731536865,
      "learning_rate": 4.947697839379778e-06,
      "loss": 0.4446,
      "step": 3450
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0881595611572266,
      "learning_rate": 4.945265275764355e-06,
      "loss": 0.71,
      "step": 3451
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.409152626991272,
      "learning_rate": 4.942832725105831e-06,
      "loss": 0.5152,
      "step": 3452
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0027379989624023,
      "learning_rate": 4.940400187980045e-06,
      "loss": 0.4631,
      "step": 3453
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2676643133163452,
      "learning_rate": 4.937967664962831e-06,
      "loss": 0.469,
      "step": 3454
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0330474376678467,
      "learning_rate": 4.93553515663002e-06,
      "loss": 0.4281,
      "step": 3455
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.389987587928772,
      "learning_rate": 4.9331026635574394e-06,
      "loss": 0.4869,
      "step": 3456
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4848506450653076,
      "learning_rate": 4.93067018632091e-06,
      "loss": 0.4684,
      "step": 3457
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2977240085601807,
      "learning_rate": 4.9282377254962606e-06,
      "loss": 0.4801,
      "step": 3458
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.108949065208435,
      "learning_rate": 4.9258052816593e-06,
      "loss": 0.5203,
      "step": 3459
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1815706491470337,
      "learning_rate": 4.9233728553858435e-06,
      "loss": 0.4978,
      "step": 3460
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.175037145614624,
      "learning_rate": 4.920940447251699e-06,
      "loss": 0.4715,
      "step": 3461
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1232078075408936,
      "learning_rate": 4.91850805783267e-06,
      "loss": 0.718,
      "step": 3462
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1009987592697144,
      "learning_rate": 4.916075687704556e-06,
      "loss": 0.4491,
      "step": 3463
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.144569993019104,
      "learning_rate": 4.913643337443152e-06,
      "loss": 0.4214,
      "step": 3464
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.280683159828186,
      "learning_rate": 4.911211007624249e-06,
      "loss": 0.5464,
      "step": 3465
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.127701759338379,
      "learning_rate": 4.908778698823633e-06,
      "loss": 0.5382,
      "step": 3466
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2699611186981201,
      "learning_rate": 4.906346411617081e-06,
      "loss": 0.5274,
      "step": 3467
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4624102115631104,
      "learning_rate": 4.903914146580368e-06,
      "loss": 0.4428,
      "step": 3468
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2935904264450073,
      "learning_rate": 4.90148190428927e-06,
      "loss": 0.4721,
      "step": 3469
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3338533639907837,
      "learning_rate": 4.899049685319549e-06,
      "loss": 0.5291,
      "step": 3470
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.2539567947387695,
      "learning_rate": 4.896617490246961e-06,
      "loss": 0.5418,
      "step": 3471
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.767867922782898,
      "learning_rate": 4.894185319647263e-06,
      "loss": 0.5042,
      "step": 3472
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0789449214935303,
      "learning_rate": 4.8917531740962e-06,
      "loss": 0.4861,
      "step": 3473
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.229739785194397,
      "learning_rate": 4.889321054169516e-06,
      "loss": 0.4609,
      "step": 3474
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2266111373901367,
      "learning_rate": 4.886888960442946e-06,
      "loss": 0.5324,
      "step": 3475
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.369145154953003,
      "learning_rate": 4.884456893492219e-06,
      "loss": 0.4653,
      "step": 3476
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.46640145778656,
      "learning_rate": 4.8820248538930585e-06,
      "loss": 0.4477,
      "step": 3477
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4872825145721436,
      "learning_rate": 4.879592842221182e-06,
      "loss": 0.4244,
      "step": 3478
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.5779402256011963,
      "learning_rate": 4.877160859052296e-06,
      "loss": 0.4685,
      "step": 3479
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.14472496509552,
      "learning_rate": 4.874728904962108e-06,
      "loss": 0.4132,
      "step": 3480
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7350214719772339,
      "learning_rate": 4.872296980526313e-06,
      "loss": 0.5225,
      "step": 3481
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1678746938705444,
      "learning_rate": 4.869865086320601e-06,
      "loss": 0.4779,
      "step": 3482
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8100875616073608,
      "learning_rate": 4.867433222920654e-06,
      "loss": 0.3916,
      "step": 3483
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0976064205169678,
      "learning_rate": 4.865001390902146e-06,
      "loss": 0.458,
      "step": 3484
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9845231175422668,
      "learning_rate": 4.862569590840744e-06,
      "loss": 0.4723,
      "step": 3485
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1637392044067383,
      "learning_rate": 4.86013782331211e-06,
      "loss": 0.4909,
      "step": 3486
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.174380302429199,
      "learning_rate": 4.857706088891895e-06,
      "loss": 0.4522,
      "step": 3487
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2109122276306152,
      "learning_rate": 4.8552743881557446e-06,
      "loss": 0.4561,
      "step": 3488
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2316642999649048,
      "learning_rate": 4.852842721679293e-06,
      "loss": 0.4092,
      "step": 3489
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2314448356628418,
      "learning_rate": 4.85041109003817e-06,
      "loss": 0.5152,
      "step": 3490
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3255951404571533,
      "learning_rate": 4.847979493807997e-06,
      "loss": 0.4731,
      "step": 3491
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1443464756011963,
      "learning_rate": 4.845547933564383e-06,
      "loss": 0.4872,
      "step": 3492
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1300678253173828,
      "learning_rate": 4.843116409882933e-06,
      "loss": 0.3895,
      "step": 3493
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0322734117507935,
      "learning_rate": 4.8406849233392415e-06,
      "loss": 0.4755,
      "step": 3494
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4221315383911133,
      "learning_rate": 4.838253474508895e-06,
      "loss": 0.4554,
      "step": 3495
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5360344648361206,
      "learning_rate": 4.835822063967468e-06,
      "loss": 0.5057,
      "step": 3496
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4869651794433594,
      "learning_rate": 4.83339069229053e-06,
      "loss": 0.5193,
      "step": 3497
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4541383981704712,
      "learning_rate": 4.8309593600536375e-06,
      "loss": 0.477,
      "step": 3498
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1661351919174194,
      "learning_rate": 4.828528067832342e-06,
      "loss": 0.4738,
      "step": 3499
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0542855262756348,
      "learning_rate": 4.826096816202182e-06,
      "loss": 0.4115,
      "step": 3500
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.07431161403656,
      "learning_rate": 4.823665605738688e-06,
      "loss": 0.4686,
      "step": 3501
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3118432760238647,
      "learning_rate": 4.821234437017381e-06,
      "loss": 0.4531,
      "step": 3502
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4517089128494263,
      "learning_rate": 4.818803310613772e-06,
      "loss": 0.4778,
      "step": 3503
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2081363201141357,
      "learning_rate": 4.8163722271033595e-06,
      "loss": 0.5127,
      "step": 3504
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6030091047286987,
      "learning_rate": 4.8139411870616346e-06,
      "loss": 0.443,
      "step": 3505
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3577964305877686,
      "learning_rate": 4.811510191064078e-06,
      "loss": 0.4706,
      "step": 3506
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.0543758869171143,
      "learning_rate": 4.80907923968616e-06,
      "loss": 0.4594,
      "step": 3507
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2317577600479126,
      "learning_rate": 4.806648333503338e-06,
      "loss": 0.7264,
      "step": 3508
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8290724754333496,
      "learning_rate": 4.804217473091061e-06,
      "loss": 0.5579,
      "step": 3509
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4351742267608643,
      "learning_rate": 4.801786659024766e-06,
      "loss": 0.4713,
      "step": 3510
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3677151203155518,
      "learning_rate": 4.79935589187988e-06,
      "loss": 0.4913,
      "step": 3511
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0638048648834229,
      "learning_rate": 4.796925172231818e-06,
      "loss": 0.4761,
      "step": 3512
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9962906837463379,
      "learning_rate": 4.794494500655986e-06,
      "loss": 0.5385,
      "step": 3513
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9654853343963623,
      "learning_rate": 4.792063877727775e-06,
      "loss": 0.5266,
      "step": 3514
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.0106561183929443,
      "learning_rate": 4.789633304022567e-06,
      "loss": 0.4818,
      "step": 3515
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0757185220718384,
      "learning_rate": 4.787202780115732e-06,
      "loss": 0.3988,
      "step": 3516
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3552895784378052,
      "learning_rate": 4.784772306582627e-06,
      "loss": 0.4977,
      "step": 3517
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0928685665130615,
      "learning_rate": 4.782341883998598e-06,
      "loss": 0.46,
      "step": 3518
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2177999019622803,
      "learning_rate": 4.779911512938981e-06,
      "loss": 0.4358,
      "step": 3519
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2099381685256958,
      "learning_rate": 4.777481193979094e-06,
      "loss": 0.4717,
      "step": 3520
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1992692947387695,
      "learning_rate": 4.77505092769425e-06,
      "loss": 0.4799,
      "step": 3521
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3675593137741089,
      "learning_rate": 4.772620714659742e-06,
      "loss": 0.5495,
      "step": 3522
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.208451509475708,
      "learning_rate": 4.770190555450858e-06,
      "loss": 0.5306,
      "step": 3523
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.743186116218567,
      "learning_rate": 4.767760450642867e-06,
      "loss": 0.4362,
      "step": 3524
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6236777305603027,
      "learning_rate": 4.76533040081103e-06,
      "loss": 0.4728,
      "step": 3525
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0865561962127686,
      "learning_rate": 4.76290040653059e-06,
      "loss": 0.4857,
      "step": 3526
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1399710178375244,
      "learning_rate": 4.760470468376782e-06,
      "loss": 0.4454,
      "step": 3527
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.729079008102417,
      "learning_rate": 4.758040586924823e-06,
      "loss": 0.5165,
      "step": 3528
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9881858825683594,
      "learning_rate": 4.755610762749921e-06,
      "loss": 0.4853,
      "step": 3529
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3251653909683228,
      "learning_rate": 4.753180996427267e-06,
      "loss": 0.4546,
      "step": 3530
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.112527847290039,
      "learning_rate": 4.750751288532039e-06,
      "loss": 0.5065,
      "step": 3531
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.875304937362671,
      "learning_rate": 4.748321639639404e-06,
      "loss": 0.5523,
      "step": 3532
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.568416953086853,
      "learning_rate": 4.74589205032451e-06,
      "loss": 0.4368,
      "step": 3533
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.529571294784546,
      "learning_rate": 4.743462521162494e-06,
      "loss": 0.4542,
      "step": 3534
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3514189720153809,
      "learning_rate": 4.741033052728477e-06,
      "loss": 0.5531,
      "step": 3535
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0204267501831055,
      "learning_rate": 4.7386036455975735e-06,
      "loss": 0.5141,
      "step": 3536
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4397861957550049,
      "learning_rate": 4.73617430034487e-06,
      "loss": 0.4535,
      "step": 3537
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2438867092132568,
      "learning_rate": 4.733745017545449e-06,
      "loss": 0.5147,
      "step": 3538
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0946710109710693,
      "learning_rate": 4.731315797774373e-06,
      "loss": 0.4057,
      "step": 3539
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0215736627578735,
      "learning_rate": 4.728886641606692e-06,
      "loss": 0.459,
      "step": 3540
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1552274227142334,
      "learning_rate": 4.72645754961744e-06,
      "loss": 0.4639,
      "step": 3541
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.096770167350769,
      "learning_rate": 4.7240285223816365e-06,
      "loss": 0.4636,
      "step": 3542
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2326043844223022,
      "learning_rate": 4.721599560474283e-06,
      "loss": 0.48,
      "step": 3543
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2583692073822021,
      "learning_rate": 4.719170664470371e-06,
      "loss": 0.4413,
      "step": 3544
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3945561647415161,
      "learning_rate": 4.716741834944869e-06,
      "loss": 0.4892,
      "step": 3545
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9175390005111694,
      "learning_rate": 4.714313072472734e-06,
      "loss": 0.4386,
      "step": 3546
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4926468133926392,
      "learning_rate": 4.711884377628911e-06,
      "loss": 0.4748,
      "step": 3547
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.2430052757263184,
      "learning_rate": 4.709455750988323e-06,
      "loss": 0.4377,
      "step": 3548
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4233205318450928,
      "learning_rate": 4.707027193125877e-06,
      "loss": 0.4605,
      "step": 3549
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.270192265510559,
      "learning_rate": 4.704598704616468e-06,
      "loss": 0.446,
      "step": 3550
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4545197486877441,
      "learning_rate": 4.70217028603497e-06,
      "loss": 0.4934,
      "step": 3551
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4672400951385498,
      "learning_rate": 4.6997419379562435e-06,
      "loss": 0.4944,
      "step": 3552
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1720731258392334,
      "learning_rate": 4.69731366095513e-06,
      "loss": 0.5219,
      "step": 3553
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.111445665359497,
      "learning_rate": 4.694885455606459e-06,
      "loss": 0.4614,
      "step": 3554
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3920707702636719,
      "learning_rate": 4.692457322485035e-06,
      "loss": 0.4421,
      "step": 3555
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2305145263671875,
      "learning_rate": 4.690029262165654e-06,
      "loss": 0.4514,
      "step": 3556
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2708511352539062,
      "learning_rate": 4.687601275223085e-06,
      "loss": 0.4202,
      "step": 3557
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1614388227462769,
      "learning_rate": 4.6851733622320915e-06,
      "loss": 0.5413,
      "step": 3558
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3413249254226685,
      "learning_rate": 4.682745523767411e-06,
      "loss": 0.4596,
      "step": 3559
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.181275725364685,
      "learning_rate": 4.680317760403765e-06,
      "loss": 0.4229,
      "step": 3560
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1234939098358154,
      "learning_rate": 4.677890072715861e-06,
      "loss": 0.518,
      "step": 3561
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.742045283317566,
      "learning_rate": 4.67546246127838e-06,
      "loss": 0.5143,
      "step": 3562
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.594539165496826,
      "learning_rate": 4.673034926665994e-06,
      "loss": 0.391,
      "step": 3563
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4841965436935425,
      "learning_rate": 4.670607469453353e-06,
      "loss": 0.458,
      "step": 3564
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0499755144119263,
      "learning_rate": 4.668180090215088e-06,
      "loss": 0.4867,
      "step": 3565
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9977362751960754,
      "learning_rate": 4.665752789525813e-06,
      "loss": 0.4869,
      "step": 3566
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2752032279968262,
      "learning_rate": 4.663325567960123e-06,
      "loss": 0.4265,
      "step": 3567
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5976375341415405,
      "learning_rate": 4.660898426092592e-06,
      "loss": 0.466,
      "step": 3568
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3043824434280396,
      "learning_rate": 4.658471364497781e-06,
      "loss": 0.4604,
      "step": 3569
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.301921367645264,
      "learning_rate": 4.656044383750227e-06,
      "loss": 0.4927,
      "step": 3570
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6257954835891724,
      "learning_rate": 4.653617484424448e-06,
      "loss": 0.4908,
      "step": 3571
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2698676586151123,
      "learning_rate": 4.6511906670949445e-06,
      "loss": 0.7498,
      "step": 3572
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3412731885910034,
      "learning_rate": 4.648763932336199e-06,
      "loss": 0.5044,
      "step": 3573
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1559066772460938,
      "learning_rate": 4.646337280722668e-06,
      "loss": 0.5031,
      "step": 3574
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4558600187301636,
      "learning_rate": 4.643910712828796e-06,
      "loss": 0.4709,
      "step": 3575
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1321393251419067,
      "learning_rate": 4.641484229229003e-06,
      "loss": 0.4477,
      "step": 3576
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1511467695236206,
      "learning_rate": 4.639057830497691e-06,
      "loss": 0.4513,
      "step": 3577
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1329501867294312,
      "learning_rate": 4.6366315172092405e-06,
      "loss": 0.5202,
      "step": 3578
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2563130855560303,
      "learning_rate": 4.634205289938013e-06,
      "loss": 0.39,
      "step": 3579
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1091365814208984,
      "learning_rate": 4.631779149258351e-06,
      "loss": 0.4459,
      "step": 3580
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3379852771759033,
      "learning_rate": 4.629353095744573e-06,
      "loss": 0.501,
      "step": 3581
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.211825132369995,
      "learning_rate": 4.6269271299709794e-06,
      "loss": 0.3807,
      "step": 3582
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1101833581924438,
      "learning_rate": 4.624501252511848e-06,
      "loss": 0.5051,
      "step": 3583
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5993221998214722,
      "learning_rate": 4.622075463941438e-06,
      "loss": 0.485,
      "step": 3584
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1705209016799927,
      "learning_rate": 4.619649764833987e-06,
      "loss": 0.5024,
      "step": 3585
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1240525245666504,
      "learning_rate": 4.617224155763709e-06,
      "loss": 0.4751,
      "step": 3586
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4089924097061157,
      "learning_rate": 4.614798637304798e-06,
      "loss": 0.4884,
      "step": 3587
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6472910642623901,
      "learning_rate": 4.612373210031428e-06,
      "loss": 0.5529,
      "step": 3588
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1381844282150269,
      "learning_rate": 4.609947874517751e-06,
      "loss": 0.4701,
      "step": 3589
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0563812255859375,
      "learning_rate": 4.607522631337895e-06,
      "loss": 0.4717,
      "step": 3590
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2102348804473877,
      "learning_rate": 4.6050974810659684e-06,
      "loss": 0.4071,
      "step": 3591
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2630263566970825,
      "learning_rate": 4.602672424276059e-06,
      "loss": 0.4752,
      "step": 3592
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8175795078277588,
      "learning_rate": 4.600247461542228e-06,
      "loss": 0.4093,
      "step": 3593
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.283881425857544,
      "learning_rate": 4.597822593438519e-06,
      "loss": 0.5331,
      "step": 3594
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7428169250488281,
      "learning_rate": 4.595397820538949e-06,
      "loss": 0.4953,
      "step": 3595
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7615461349487305,
      "learning_rate": 4.592973143417515e-06,
      "loss": 0.4504,
      "step": 3596
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5342744588851929,
      "learning_rate": 4.59054856264819e-06,
      "loss": 0.4887,
      "step": 3597
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2367608547210693,
      "learning_rate": 4.588124078804929e-06,
      "loss": 0.4551,
      "step": 3598
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1911118030548096,
      "learning_rate": 4.585699692461655e-06,
      "loss": 0.7239,
      "step": 3599
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.613885760307312,
      "learning_rate": 4.5832754041922764e-06,
      "loss": 0.441,
      "step": 3600
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2106894254684448,
      "learning_rate": 4.580851214570672e-06,
      "loss": 0.7083,
      "step": 3601
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2221572399139404,
      "learning_rate": 4.5784271241707015e-06,
      "loss": 0.7145,
      "step": 3602
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7298587560653687,
      "learning_rate": 4.576003133566202e-06,
      "loss": 0.4985,
      "step": 3603
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2017384767532349,
      "learning_rate": 4.573579243330982e-06,
      "loss": 0.4047,
      "step": 3604
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4308403730392456,
      "learning_rate": 4.5711554540388305e-06,
      "loss": 0.4819,
      "step": 3605
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9699112176895142,
      "learning_rate": 4.568731766263511e-06,
      "loss": 0.5249,
      "step": 3606
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.197013258934021,
      "learning_rate": 4.566308180578763e-06,
      "loss": 0.5036,
      "step": 3607
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.555896043777466,
      "learning_rate": 4.563884697558301e-06,
      "loss": 0.5093,
      "step": 3608
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2863274812698364,
      "learning_rate": 4.561461317775817e-06,
      "loss": 0.5036,
      "step": 3609
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3041518926620483,
      "learning_rate": 4.559038041804979e-06,
      "loss": 0.4937,
      "step": 3610
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5069419145584106,
      "learning_rate": 4.556614870219426e-06,
      "loss": 0.4427,
      "step": 3611
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6954445838928223,
      "learning_rate": 4.554191803592778e-06,
      "loss": 0.4738,
      "step": 3612
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2496169805526733,
      "learning_rate": 4.551768842498623e-06,
      "loss": 0.4666,
      "step": 3613
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3874828815460205,
      "learning_rate": 4.549345987510537e-06,
      "loss": 0.45,
      "step": 3614
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3682392835617065,
      "learning_rate": 4.546923239202057e-06,
      "loss": 0.474,
      "step": 3615
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1592888832092285,
      "learning_rate": 4.5445005981467e-06,
      "loss": 0.4267,
      "step": 3616
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0302428007125854,
      "learning_rate": 4.542078064917959e-06,
      "loss": 0.4944,
      "step": 3617
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9875468611717224,
      "learning_rate": 4.539655640089301e-06,
      "loss": 0.4949,
      "step": 3618
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3191542625427246,
      "learning_rate": 4.5372333242341645e-06,
      "loss": 0.455,
      "step": 3619
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1110254526138306,
      "learning_rate": 4.534811117925967e-06,
      "loss": 0.3858,
      "step": 3620
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2800278663635254,
      "learning_rate": 4.532389021738095e-06,
      "loss": 0.5095,
      "step": 3621
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5259616374969482,
      "learning_rate": 4.529967036243914e-06,
      "loss": 0.4897,
      "step": 3622
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1389999389648438,
      "learning_rate": 4.527545162016758e-06,
      "loss": 0.5191,
      "step": 3623
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2904798984527588,
      "learning_rate": 4.525123399629936e-06,
      "loss": 0.4393,
      "step": 3624
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1469279527664185,
      "learning_rate": 4.522701749656736e-06,
      "loss": 0.4265,
      "step": 3625
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.05861234664917,
      "learning_rate": 4.520280212670414e-06,
      "loss": 0.4842,
      "step": 3626
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.885891914367676,
      "learning_rate": 4.517858789244202e-06,
      "loss": 0.5526,
      "step": 3627
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1371822357177734,
      "learning_rate": 4.515437479951299e-06,
      "loss": 0.4818,
      "step": 3628
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0891015529632568,
      "learning_rate": 4.513016285364884e-06,
      "loss": 0.4619,
      "step": 3629
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1517295837402344,
      "learning_rate": 4.510595206058107e-06,
      "loss": 0.4467,
      "step": 3630
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2049331665039062,
      "learning_rate": 4.508174242604089e-06,
      "loss": 0.4858,
      "step": 3631
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2446701526641846,
      "learning_rate": 4.505753395575926e-06,
      "loss": 0.4951,
      "step": 3632
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9932177662849426,
      "learning_rate": 4.503332665546684e-06,
      "loss": 0.3979,
      "step": 3633
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4116859436035156,
      "learning_rate": 4.500912053089404e-06,
      "loss": 0.4474,
      "step": 3634
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.66689932346344,
      "learning_rate": 4.498491558777094e-06,
      "loss": 0.511,
      "step": 3635
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2981079816818237,
      "learning_rate": 4.496071183182742e-06,
      "loss": 0.4538,
      "step": 3636
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.757336139678955,
      "learning_rate": 4.4936509268793014e-06,
      "loss": 0.3963,
      "step": 3637
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3441303968429565,
      "learning_rate": 4.4912307904397e-06,
      "loss": 0.4867,
      "step": 3638
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5726954936981201,
      "learning_rate": 4.488810774436838e-06,
      "loss": 0.5035,
      "step": 3639
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1681522130966187,
      "learning_rate": 4.486390879443583e-06,
      "loss": 0.4341,
      "step": 3640
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1052411794662476,
      "learning_rate": 4.483971106032778e-06,
      "loss": 0.4855,
      "step": 3641
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0732966661453247,
      "learning_rate": 4.4815514547772365e-06,
      "loss": 0.4434,
      "step": 3642
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2136074304580688,
      "learning_rate": 4.479131926249743e-06,
      "loss": 0.5004,
      "step": 3643
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3537875413894653,
      "learning_rate": 4.47671252102305e-06,
      "loss": 0.5606,
      "step": 3644
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0729570388793945,
      "learning_rate": 4.474293239669886e-06,
      "loss": 0.4495,
      "step": 3645
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1562659740447998,
      "learning_rate": 4.4718740827629456e-06,
      "loss": 0.4632,
      "step": 3646
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.263038158416748,
      "learning_rate": 4.469455050874898e-06,
      "loss": 0.495,
      "step": 3647
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7173744440078735,
      "learning_rate": 4.467036144578381e-06,
      "loss": 0.4746,
      "step": 3648
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0812791585922241,
      "learning_rate": 4.464617364445999e-06,
      "loss": 0.4932,
      "step": 3649
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2158037424087524,
      "learning_rate": 4.462198711050333e-06,
      "loss": 0.4406,
      "step": 3650
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0282524824142456,
      "learning_rate": 4.45978018496393e-06,
      "loss": 0.4264,
      "step": 3651
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2747416496276855,
      "learning_rate": 4.457361786759306e-06,
      "loss": 0.5161,
      "step": 3652
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.339534044265747,
      "learning_rate": 4.45494351700895e-06,
      "loss": 0.4973,
      "step": 3653
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2130005359649658,
      "learning_rate": 4.452525376285319e-06,
      "loss": 0.4122,
      "step": 3654
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0633167028427124,
      "learning_rate": 4.450107365160839e-06,
      "loss": 0.4355,
      "step": 3655
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.7850449085235596,
      "learning_rate": 4.447689484207907e-06,
      "loss": 0.4437,
      "step": 3656
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1346746683120728,
      "learning_rate": 4.445271733998886e-06,
      "loss": 0.5427,
      "step": 3657
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1173310279846191,
      "learning_rate": 4.442854115106112e-06,
      "loss": 0.4803,
      "step": 3658
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2775988578796387,
      "learning_rate": 4.440436628101887e-06,
      "loss": 0.4228,
      "step": 3659
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1961290836334229,
      "learning_rate": 4.438019273558484e-06,
      "loss": 0.5274,
      "step": 3660
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0616527795791626,
      "learning_rate": 4.435602052048142e-06,
      "loss": 0.4336,
      "step": 3661
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.36309015750885,
      "learning_rate": 4.43318496414307e-06,
      "loss": 0.5636,
      "step": 3662
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2391799688339233,
      "learning_rate": 4.4307680104154466e-06,
      "loss": 0.4953,
      "step": 3663
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1975343227386475,
      "learning_rate": 4.428351191437416e-06,
      "loss": 0.4874,
      "step": 3664
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4796069860458374,
      "learning_rate": 4.425934507781091e-06,
      "loss": 0.5389,
      "step": 3665
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.158341884613037,
      "learning_rate": 4.423517960018555e-06,
      "loss": 0.7276,
      "step": 3666
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2984181642532349,
      "learning_rate": 4.421101548721857e-06,
      "loss": 0.4766,
      "step": 3667
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.527794361114502,
      "learning_rate": 4.418685274463012e-06,
      "loss": 0.4953,
      "step": 3668
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0827174186706543,
      "learning_rate": 4.4162691378140085e-06,
      "loss": 0.4829,
      "step": 3669
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5772262811660767,
      "learning_rate": 4.413853139346796e-06,
      "loss": 0.4508,
      "step": 3670
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.337611436843872,
      "learning_rate": 4.411437279633295e-06,
      "loss": 0.4763,
      "step": 3671
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9402848482131958,
      "learning_rate": 4.409021559245391e-06,
      "loss": 0.4753,
      "step": 3672
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.355582594871521,
      "learning_rate": 4.406605978754939e-06,
      "loss": 0.4701,
      "step": 3673
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.071126937866211,
      "learning_rate": 4.404190538733758e-06,
      "loss": 0.4988,
      "step": 3674
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.040247917175293,
      "learning_rate": 4.401775239753635e-06,
      "loss": 0.4294,
      "step": 3675
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.567387580871582,
      "learning_rate": 4.399360082386326e-06,
      "loss": 0.4907,
      "step": 3676
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.05764901638031,
      "learning_rate": 4.39694506720355e-06,
      "loss": 0.5147,
      "step": 3677
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2167812585830688,
      "learning_rate": 4.394530194776992e-06,
      "loss": 0.4,
      "step": 3678
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1195634603500366,
      "learning_rate": 4.392115465678305e-06,
      "loss": 0.4892,
      "step": 3679
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2798612117767334,
      "learning_rate": 4.389700880479112e-06,
      "loss": 0.7517,
      "step": 3680
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.138463258743286,
      "learning_rate": 4.387286439750994e-06,
      "loss": 0.4514,
      "step": 3681
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0792973041534424,
      "learning_rate": 4.384872144065503e-06,
      "loss": 0.5254,
      "step": 3682
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5382187366485596,
      "learning_rate": 4.3824579939941545e-06,
      "loss": 0.4876,
      "step": 3683
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3319951295852661,
      "learning_rate": 4.3800439901084305e-06,
      "loss": 0.4553,
      "step": 3684
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.158719062805176,
      "learning_rate": 4.377630132979779e-06,
      "loss": 0.5421,
      "step": 3685
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2249597311019897,
      "learning_rate": 4.375216423179612e-06,
      "loss": 0.5389,
      "step": 3686
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.21818208694458,
      "learning_rate": 4.372802861279307e-06,
      "loss": 0.4926,
      "step": 3687
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.375298261642456,
      "learning_rate": 4.370389447850208e-06,
      "loss": 0.4398,
      "step": 3688
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9941698312759399,
      "learning_rate": 4.36797618346362e-06,
      "loss": 0.4338,
      "step": 3689
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5074272155761719,
      "learning_rate": 4.365563068690816e-06,
      "loss": 0.4571,
      "step": 3690
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.129909873008728,
      "learning_rate": 4.363150104103031e-06,
      "loss": 0.3658,
      "step": 3691
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8398222923278809,
      "learning_rate": 4.3607372902714715e-06,
      "loss": 0.4077,
      "step": 3692
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2031002044677734,
      "learning_rate": 4.358324627767298e-06,
      "loss": 0.449,
      "step": 3693
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3843092918395996,
      "learning_rate": 4.355912117161643e-06,
      "loss": 0.4626,
      "step": 3694
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0304641723632812,
      "learning_rate": 4.3534997590256e-06,
      "loss": 0.4424,
      "step": 3695
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3214027881622314,
      "learning_rate": 4.351087553930224e-06,
      "loss": 0.4968,
      "step": 3696
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2093092203140259,
      "learning_rate": 4.3486755024465395e-06,
      "loss": 0.4245,
      "step": 3697
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2019516229629517,
      "learning_rate": 4.346263605145531e-06,
      "loss": 0.4775,
      "step": 3698
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2362747192382812,
      "learning_rate": 4.343851862598144e-06,
      "loss": 0.4369,
      "step": 3699
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8305219411849976,
      "learning_rate": 4.341440275375296e-06,
      "loss": 0.5079,
      "step": 3700
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.059851884841919,
      "learning_rate": 4.339028844047858e-06,
      "loss": 0.5,
      "step": 3701
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1858259439468384,
      "learning_rate": 4.3366175691866655e-06,
      "loss": 0.5171,
      "step": 3702
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.384266972541809,
      "learning_rate": 4.334206451362526e-06,
      "loss": 0.5173,
      "step": 3703
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1801835298538208,
      "learning_rate": 4.331795491146201e-06,
      "loss": 0.7107,
      "step": 3704
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2409790754318237,
      "learning_rate": 4.329384689108419e-06,
      "loss": 0.4639,
      "step": 3705
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1232680082321167,
      "learning_rate": 4.326974045819865e-06,
      "loss": 0.5231,
      "step": 3706
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.514146089553833,
      "learning_rate": 4.324563561851195e-06,
      "loss": 0.4798,
      "step": 3707
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.428574562072754,
      "learning_rate": 4.322153237773019e-06,
      "loss": 0.5285,
      "step": 3708
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.063516616821289,
      "learning_rate": 4.319743074155916e-06,
      "loss": 0.443,
      "step": 3709
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.36549973487854,
      "learning_rate": 4.3173330715704245e-06,
      "loss": 0.5106,
      "step": 3710
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2263104915618896,
      "learning_rate": 4.314923230587043e-06,
      "loss": 0.5427,
      "step": 3711
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.580024003982544,
      "learning_rate": 4.312513551776233e-06,
      "loss": 0.5016,
      "step": 3712
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2963329553604126,
      "learning_rate": 4.310104035708418e-06,
      "loss": 0.4133,
      "step": 3713
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2493406534194946,
      "learning_rate": 4.307694682953984e-06,
      "loss": 0.4093,
      "step": 3714
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2739354372024536,
      "learning_rate": 4.305285494083277e-06,
      "loss": 0.415,
      "step": 3715
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1892449855804443,
      "learning_rate": 4.302876469666604e-06,
      "loss": 0.4339,
      "step": 3716
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.494749665260315,
      "learning_rate": 4.300467610274234e-06,
      "loss": 0.4986,
      "step": 3717
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1635042428970337,
      "learning_rate": 4.298058916476397e-06,
      "loss": 0.4927,
      "step": 3718
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2522603273391724,
      "learning_rate": 4.29565038884328e-06,
      "loss": 0.4303,
      "step": 3719
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4066143035888672,
      "learning_rate": 4.293242027945036e-06,
      "loss": 0.4134,
      "step": 3720
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0285311937332153,
      "learning_rate": 4.290833834351778e-06,
      "loss": 0.4298,
      "step": 3721
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.384130835533142,
      "learning_rate": 4.2884258086335755e-06,
      "loss": 0.4738,
      "step": 3722
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2785758972167969,
      "learning_rate": 4.286017951360461e-06,
      "loss": 0.4748,
      "step": 3723
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4369633197784424,
      "learning_rate": 4.283610263102427e-06,
      "loss": 0.4284,
      "step": 3724
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7088035345077515,
      "learning_rate": 4.281202744429426e-06,
      "loss": 0.442,
      "step": 3725
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.151557445526123,
      "learning_rate": 4.278795395911369e-06,
      "loss": 0.4298,
      "step": 3726
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2916895151138306,
      "learning_rate": 4.276388218118129e-06,
      "loss": 0.4814,
      "step": 3727
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0216703414916992,
      "learning_rate": 4.2739812116195375e-06,
      "loss": 0.6988,
      "step": 3728
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0324569940567017,
      "learning_rate": 4.271574376985386e-06,
      "loss": 0.4676,
      "step": 3729
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.597609043121338,
      "learning_rate": 4.269167714785421e-06,
      "loss": 0.4796,
      "step": 3730
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3925524950027466,
      "learning_rate": 4.266761225589354e-06,
      "loss": 0.4494,
      "step": 3731
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.25184166431427,
      "learning_rate": 4.264354909966854e-06,
      "loss": 0.7518,
      "step": 3732
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.31006920337677,
      "learning_rate": 4.261948768487547e-06,
      "loss": 0.747,
      "step": 3733
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4337772130966187,
      "learning_rate": 4.259542801721021e-06,
      "loss": 0.4657,
      "step": 3734
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5485398769378662,
      "learning_rate": 4.2571370102368165e-06,
      "loss": 0.4575,
      "step": 3735
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1060177087783813,
      "learning_rate": 4.254731394604441e-06,
      "loss": 0.4645,
      "step": 3736
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6442610025405884,
      "learning_rate": 4.252325955393355e-06,
      "loss": 0.4408,
      "step": 3737
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1385222673416138,
      "learning_rate": 4.249920693172978e-06,
      "loss": 0.4549,
      "step": 3738
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.398979663848877,
      "learning_rate": 4.2475156085126865e-06,
      "loss": 0.411,
      "step": 3739
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4283634424209595,
      "learning_rate": 4.245110701981818e-06,
      "loss": 0.5425,
      "step": 3740
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2237149477005005,
      "learning_rate": 4.242705974149666e-06,
      "loss": 0.5044,
      "step": 3741
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1937615871429443,
      "learning_rate": 4.240301425585483e-06,
      "loss": 0.7365,
      "step": 3742
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2112138271331787,
      "learning_rate": 4.237897056858475e-06,
      "loss": 0.4892,
      "step": 3743
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2130614519119263,
      "learning_rate": 4.23549286853781e-06,
      "loss": 0.4652,
      "step": 3744
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2690411806106567,
      "learning_rate": 4.233088861192611e-06,
      "loss": 0.4238,
      "step": 3745
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.512536644935608,
      "learning_rate": 4.2306850353919594e-06,
      "loss": 0.5116,
      "step": 3746
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1801817417144775,
      "learning_rate": 4.228281391704894e-06,
      "loss": 0.4232,
      "step": 3747
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1670353412628174,
      "learning_rate": 4.225877930700408e-06,
      "loss": 0.433,
      "step": 3748
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.587514877319336,
      "learning_rate": 4.223474652947453e-06,
      "loss": 0.4747,
      "step": 3749
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.297468900680542,
      "learning_rate": 4.221071559014939e-06,
      "loss": 0.4438,
      "step": 3750
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3069024085998535,
      "learning_rate": 4.218668649471728e-06,
      "loss": 0.4703,
      "step": 3751
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3580825328826904,
      "learning_rate": 4.216265924886642e-06,
      "loss": 0.5356,
      "step": 3752
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.66928768157959,
      "learning_rate": 4.213863385828459e-06,
      "loss": 0.5042,
      "step": 3753
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2629647254943848,
      "learning_rate": 4.211461032865914e-06,
      "loss": 0.4551,
      "step": 3754
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5593141317367554,
      "learning_rate": 4.2090588665676905e-06,
      "loss": 0.5098,
      "step": 3755
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1978681087493896,
      "learning_rate": 4.206656887502437e-06,
      "loss": 0.4754,
      "step": 3756
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.243452787399292,
      "learning_rate": 4.204255096238753e-06,
      "loss": 0.5013,
      "step": 3757
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.108431339263916,
      "learning_rate": 4.201853493345198e-06,
      "loss": 0.452,
      "step": 3758
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3630956411361694,
      "learning_rate": 4.199452079390281e-06,
      "loss": 0.4636,
      "step": 3759
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1597799062728882,
      "learning_rate": 4.197050854942469e-06,
      "loss": 0.5245,
      "step": 3760
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.04667592048645,
      "learning_rate": 4.194649820570184e-06,
      "loss": 0.4384,
      "step": 3761
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6743823289871216,
      "learning_rate": 4.192248976841804e-06,
      "loss": 0.4623,
      "step": 3762
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.425136685371399,
      "learning_rate": 4.18984832432566e-06,
      "loss": 0.4319,
      "step": 3763
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5944958925247192,
      "learning_rate": 4.187447863590039e-06,
      "loss": 0.4462,
      "step": 3764
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3894093036651611,
      "learning_rate": 4.185047595203183e-06,
      "loss": 0.4827,
      "step": 3765
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.252699851989746,
      "learning_rate": 4.182647519733286e-06,
      "loss": 0.4595,
      "step": 3766
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9561468958854675,
      "learning_rate": 4.1802476377484994e-06,
      "loss": 0.4254,
      "step": 3767
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2130757570266724,
      "learning_rate": 4.177847949816927e-06,
      "loss": 0.5082,
      "step": 3768
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.241726875305176,
      "learning_rate": 4.175448456506626e-06,
      "loss": 0.5458,
      "step": 3769
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1178290843963623,
      "learning_rate": 4.17304915838561e-06,
      "loss": 0.4288,
      "step": 3770
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.181067705154419,
      "learning_rate": 4.170650056021847e-06,
      "loss": 0.7302,
      "step": 3771
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6164624691009521,
      "learning_rate": 4.168251149983255e-06,
      "loss": 0.5895,
      "step": 3772
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.025567650794983,
      "learning_rate": 4.165852440837705e-06,
      "loss": 0.4514,
      "step": 3773
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.218166708946228,
      "learning_rate": 4.163453929153026e-06,
      "loss": 0.5101,
      "step": 3774
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4892791509628296,
      "learning_rate": 4.161055615496999e-06,
      "loss": 0.5171,
      "step": 3775
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6306260824203491,
      "learning_rate": 4.158657500437353e-06,
      "loss": 0.4955,
      "step": 3776
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6776310205459595,
      "learning_rate": 4.156259584541778e-06,
      "loss": 0.5257,
      "step": 3777
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.615678310394287,
      "learning_rate": 4.153861868377912e-06,
      "loss": 0.4165,
      "step": 3778
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.166347861289978,
      "learning_rate": 4.151464352513344e-06,
      "loss": 0.486,
      "step": 3779
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2497451305389404,
      "learning_rate": 4.149067037515618e-06,
      "loss": 0.4927,
      "step": 3780
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.070997953414917,
      "learning_rate": 4.146669923952236e-06,
      "loss": 0.5164,
      "step": 3781
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.328606128692627,
      "learning_rate": 4.1442730123906425e-06,
      "loss": 0.4951,
      "step": 3782
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2314821481704712,
      "learning_rate": 4.141876303398241e-06,
      "loss": 0.5067,
      "step": 3783
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0695781707763672,
      "learning_rate": 4.1394797975423825e-06,
      "loss": 0.4443,
      "step": 3784
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5186100006103516,
      "learning_rate": 4.137083495390372e-06,
      "loss": 0.4625,
      "step": 3785
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4401565790176392,
      "learning_rate": 4.134687397509468e-06,
      "loss": 0.3908,
      "step": 3786
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4085795879364014,
      "learning_rate": 4.132291504466878e-06,
      "loss": 0.4615,
      "step": 3787
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1851587295532227,
      "learning_rate": 4.129895816829761e-06,
      "loss": 0.512,
      "step": 3788
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3753005266189575,
      "learning_rate": 4.127500335165231e-06,
      "loss": 0.4265,
      "step": 3789
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.372316837310791,
      "learning_rate": 4.125105060040348e-06,
      "loss": 0.4063,
      "step": 3790
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3908714056015015,
      "learning_rate": 4.122709992022127e-06,
      "loss": 0.492,
      "step": 3791
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.1996452808380127,
      "learning_rate": 4.1203151316775324e-06,
      "loss": 0.4878,
      "step": 3792
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0662308931350708,
      "learning_rate": 4.1179204795734805e-06,
      "loss": 0.4083,
      "step": 3793
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2699750661849976,
      "learning_rate": 4.1155260362768364e-06,
      "loss": 0.5407,
      "step": 3794
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5860843658447266,
      "learning_rate": 4.11313180235442e-06,
      "loss": 0.4981,
      "step": 3795
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4819947481155396,
      "learning_rate": 4.110737778372993e-06,
      "loss": 0.5228,
      "step": 3796
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.189831256866455,
      "learning_rate": 4.108343964899277e-06,
      "loss": 0.754,
      "step": 3797
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3608757257461548,
      "learning_rate": 4.105950362499939e-06,
      "loss": 0.4952,
      "step": 3798
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2097973823547363,
      "learning_rate": 4.103556971741596e-06,
      "loss": 0.4939,
      "step": 3799
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.5768046379089355,
      "learning_rate": 4.101163793190815e-06,
      "loss": 0.5003,
      "step": 3800
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4472863674163818,
      "learning_rate": 4.098770827414116e-06,
      "loss": 0.4652,
      "step": 3801
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.369110345840454,
      "learning_rate": 4.096378074977962e-06,
      "loss": 0.4529,
      "step": 3802
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3834346532821655,
      "learning_rate": 4.093985536448774e-06,
      "loss": 0.4732,
      "step": 3803
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7142869234085083,
      "learning_rate": 4.091593212392916e-06,
      "loss": 0.4171,
      "step": 3804
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.717590093612671,
      "learning_rate": 4.0892011033767035e-06,
      "loss": 0.4886,
      "step": 3805
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4305800199508667,
      "learning_rate": 4.086809209966399e-06,
      "loss": 0.523,
      "step": 3806
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7468712329864502,
      "learning_rate": 4.0844175327282195e-06,
      "loss": 0.4687,
      "step": 3807
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8088017702102661,
      "learning_rate": 4.0820260722283226e-06,
      "loss": 0.4662,
      "step": 3808
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4492945671081543,
      "learning_rate": 4.079634829032821e-06,
      "loss": 0.4994,
      "step": 3809
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4150289297103882,
      "learning_rate": 4.077243803707774e-06,
      "loss": 0.5121,
      "step": 3810
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2162574529647827,
      "learning_rate": 4.07485299681919e-06,
      "loss": 0.5298,
      "step": 3811
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.450033187866211,
      "learning_rate": 4.072462408933023e-06,
      "loss": 0.39,
      "step": 3812
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3646560907363892,
      "learning_rate": 4.0700720406151784e-06,
      "loss": 0.4541,
      "step": 3813
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1552796363830566,
      "learning_rate": 4.067681892431509e-06,
      "loss": 0.4988,
      "step": 3814
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.233593463897705,
      "learning_rate": 4.065291964947814e-06,
      "loss": 0.4857,
      "step": 3815
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1459215879440308,
      "learning_rate": 4.062902258729842e-06,
      "loss": 0.4038,
      "step": 3816
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4982690811157227,
      "learning_rate": 4.060512774343287e-06,
      "loss": 0.5141,
      "step": 3817
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2073893547058105,
      "learning_rate": 4.058123512353793e-06,
      "loss": 0.5351,
      "step": 3818
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0143953561782837,
      "learning_rate": 4.0557344733269505e-06,
      "loss": 0.4955,
      "step": 3819
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1365998983383179,
      "learning_rate": 4.053345657828298e-06,
      "loss": 0.5165,
      "step": 3820
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.334534764289856,
      "learning_rate": 4.050957066423316e-06,
      "loss": 0.4685,
      "step": 3821
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9485896229743958,
      "learning_rate": 4.048568699677441e-06,
      "loss": 0.3771,
      "step": 3822
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5398179292678833,
      "learning_rate": 4.0461805581560475e-06,
      "loss": 0.4895,
      "step": 3823
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2216392755508423,
      "learning_rate": 4.043792642424461e-06,
      "loss": 0.4908,
      "step": 3824
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2885332107543945,
      "learning_rate": 4.041404953047957e-06,
      "loss": 0.3928,
      "step": 3825
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3729033470153809,
      "learning_rate": 4.0390174905917486e-06,
      "loss": 0.465,
      "step": 3826
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1820400953292847,
      "learning_rate": 4.036630255621004e-06,
      "loss": 0.5271,
      "step": 3827
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.340897560119629,
      "learning_rate": 4.0342432487008304e-06,
      "loss": 0.477,
      "step": 3828
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1084380149841309,
      "learning_rate": 4.031856470396286e-06,
      "loss": 0.7476,
      "step": 3829
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1119016408920288,
      "learning_rate": 4.029469921272373e-06,
      "loss": 0.4929,
      "step": 3830
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6941461563110352,
      "learning_rate": 4.02708360189404e-06,
      "loss": 0.4785,
      "step": 3831
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8331741094589233,
      "learning_rate": 4.02469751282618e-06,
      "loss": 0.4693,
      "step": 3832
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4400092363357544,
      "learning_rate": 4.022311654633631e-06,
      "loss": 0.4449,
      "step": 3833
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.334692120552063,
      "learning_rate": 4.01992602788118e-06,
      "loss": 0.4616,
      "step": 3834
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1128520965576172,
      "learning_rate": 4.017540633133553e-06,
      "loss": 0.7295,
      "step": 3835
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.5996947288513184,
      "learning_rate": 4.01515547095543e-06,
      "loss": 0.4986,
      "step": 3836
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4900867938995361,
      "learning_rate": 4.0127705419114275e-06,
      "loss": 0.4736,
      "step": 3837
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4011931419372559,
      "learning_rate": 4.0103858465661105e-06,
      "loss": 0.5012,
      "step": 3838
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2217152118682861,
      "learning_rate": 4.0080013854839885e-06,
      "loss": 0.4742,
      "step": 3839
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2493711709976196,
      "learning_rate": 4.005617159229515e-06,
      "loss": 0.4999,
      "step": 3840
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1532831192016602,
      "learning_rate": 4.003233168367088e-06,
      "loss": 0.5643,
      "step": 3841
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3443459272384644,
      "learning_rate": 4.000849413461051e-06,
      "loss": 0.4699,
      "step": 3842
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.223336935043335,
      "learning_rate": 3.9984658950756896e-06,
      "loss": 0.4937,
      "step": 3843
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1362165212631226,
      "learning_rate": 3.996082613775236e-06,
      "loss": 0.4608,
      "step": 3844
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9008657932281494,
      "learning_rate": 3.993699570123864e-06,
      "loss": 0.4825,
      "step": 3845
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.49198579788208,
      "learning_rate": 3.991316764685687e-06,
      "loss": 0.5338,
      "step": 3846
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4034321308135986,
      "learning_rate": 3.988934198024775e-06,
      "loss": 0.4808,
      "step": 3847
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.15071702003479,
      "learning_rate": 3.986551870705129e-06,
      "loss": 0.4831,
      "step": 3848
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0583370923995972,
      "learning_rate": 3.9841697832907e-06,
      "loss": 0.4864,
      "step": 3849
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.261181354522705,
      "learning_rate": 3.981787936345377e-06,
      "loss": 0.5079,
      "step": 3850
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3590184450149536,
      "learning_rate": 3.979406330432996e-06,
      "loss": 0.4306,
      "step": 3851
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3951691389083862,
      "learning_rate": 3.977024966117336e-06,
      "loss": 0.3962,
      "step": 3852
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5513807535171509,
      "learning_rate": 3.974643843962116e-06,
      "loss": 0.467,
      "step": 3853
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4080134630203247,
      "learning_rate": 3.972262964531001e-06,
      "loss": 0.4176,
      "step": 3854
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.177771806716919,
      "learning_rate": 3.969882328387596e-06,
      "loss": 0.4489,
      "step": 3855
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2481471300125122,
      "learning_rate": 3.96750193609545e-06,
      "loss": 0.4598,
      "step": 3856
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5584619045257568,
      "learning_rate": 3.965121788218054e-06,
      "loss": 0.4607,
      "step": 3857
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3126916885375977,
      "learning_rate": 3.962741885318836e-06,
      "loss": 0.4536,
      "step": 3858
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8074747323989868,
      "learning_rate": 3.960362227961178e-06,
      "loss": 0.4922,
      "step": 3859
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2481772899627686,
      "learning_rate": 3.9579828167083935e-06,
      "loss": 0.4059,
      "step": 3860
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3922150135040283,
      "learning_rate": 3.955603652123741e-06,
      "loss": 0.442,
      "step": 3861
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0421425104141235,
      "learning_rate": 3.95322473477042e-06,
      "loss": 0.7111,
      "step": 3862
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4324030876159668,
      "learning_rate": 3.950846065211573e-06,
      "loss": 0.4799,
      "step": 3863
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1093989610671997,
      "learning_rate": 3.948467644010282e-06,
      "loss": 0.7078,
      "step": 3864
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0970591306686401,
      "learning_rate": 3.9460894717295705e-06,
      "loss": 0.4695,
      "step": 3865
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2620539665222168,
      "learning_rate": 3.9437115489324045e-06,
      "loss": 0.4672,
      "step": 3866
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6384423971176147,
      "learning_rate": 3.94133387618169e-06,
      "loss": 0.5068,
      "step": 3867
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5573164224624634,
      "learning_rate": 3.9389564540402736e-06,
      "loss": 0.4978,
      "step": 3868
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.372050166130066,
      "learning_rate": 3.9365792830709425e-06,
      "loss": 0.5807,
      "step": 3869
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.486539363861084,
      "learning_rate": 3.934202363836426e-06,
      "loss": 0.5206,
      "step": 3870
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3325936794281006,
      "learning_rate": 3.931825696899391e-06,
      "loss": 0.4789,
      "step": 3871
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.428023338317871,
      "learning_rate": 3.929449282822448e-06,
      "loss": 0.4943,
      "step": 3872
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2595998048782349,
      "learning_rate": 3.927073122168145e-06,
      "loss": 0.5121,
      "step": 3873
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.175004005432129,
      "learning_rate": 3.924697215498971e-06,
      "loss": 0.4947,
      "step": 3874
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1975024938583374,
      "learning_rate": 3.922321563377354e-06,
      "loss": 0.4744,
      "step": 3875
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1722044944763184,
      "learning_rate": 3.919946166365664e-06,
      "loss": 0.4906,
      "step": 3876
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.093692660331726,
      "learning_rate": 3.917571025026207e-06,
      "loss": 0.3979,
      "step": 3877
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6676005125045776,
      "learning_rate": 3.915196139921233e-06,
      "loss": 0.4949,
      "step": 3878
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3735876083374023,
      "learning_rate": 3.912821511612927e-06,
      "loss": 0.4395,
      "step": 3879
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.631836414337158,
      "learning_rate": 3.910447140663416e-06,
      "loss": 0.4726,
      "step": 3880
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5191874504089355,
      "learning_rate": 3.908073027634765e-06,
      "loss": 0.4791,
      "step": 3881
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3812601566314697,
      "learning_rate": 3.90569917308898e-06,
      "loss": 0.4831,
      "step": 3882
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1406773328781128,
      "learning_rate": 3.903325577588002e-06,
      "loss": 0.4757,
      "step": 3883
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7666069269180298,
      "learning_rate": 3.9009522416937135e-06,
      "loss": 0.4323,
      "step": 3884
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.6199862957000732,
      "learning_rate": 3.898579165967934e-06,
      "loss": 0.4972,
      "step": 3885
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.508803367614746,
      "learning_rate": 3.896206350972424e-06,
      "loss": 0.5285,
      "step": 3886
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3388100862503052,
      "learning_rate": 3.8938337972688775e-06,
      "loss": 0.4474,
      "step": 3887
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9173970222473145,
      "learning_rate": 3.891461505418931e-06,
      "loss": 0.4201,
      "step": 3888
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.9985344409942627,
      "learning_rate": 3.889089475984159e-06,
      "loss": 0.5344,
      "step": 3889
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.30409836769104,
      "learning_rate": 3.8867177095260695e-06,
      "loss": 0.4738,
      "step": 3890
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0682761669158936,
      "learning_rate": 3.8843462066061126e-06,
      "loss": 0.467,
      "step": 3891
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3618712425231934,
      "learning_rate": 3.881974967785675e-06,
      "loss": 0.4847,
      "step": 3892
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0191632509231567,
      "learning_rate": 3.879603993626081e-06,
      "loss": 0.4083,
      "step": 3893
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4770653247833252,
      "learning_rate": 3.877233284688591e-06,
      "loss": 0.4864,
      "step": 3894
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2057909965515137,
      "learning_rate": 3.874862841534403e-06,
      "loss": 0.5125,
      "step": 3895
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.1551411151885986,
      "learning_rate": 3.872492664724654e-06,
      "loss": 0.4856,
      "step": 3896
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0315207242965698,
      "learning_rate": 3.870122754820415e-06,
      "loss": 0.6734,
      "step": 3897
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2824029922485352,
      "learning_rate": 3.867753112382696e-06,
      "loss": 0.4954,
      "step": 3898
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5443288087844849,
      "learning_rate": 3.865383737972441e-06,
      "loss": 0.4942,
      "step": 3899
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8717470169067383,
      "learning_rate": 3.863014632150535e-06,
      "loss": 0.4449,
      "step": 3900
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.378735899925232,
      "learning_rate": 3.860645795477795e-06,
      "loss": 0.5339,
      "step": 3901
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.430708885192871,
      "learning_rate": 3.8582772285149765e-06,
      "loss": 0.5429,
      "step": 3902
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.426598310470581,
      "learning_rate": 3.855908931822772e-06,
      "loss": 0.4583,
      "step": 3903
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.168117046356201,
      "learning_rate": 3.8535409059618085e-06,
      "loss": 0.4955,
      "step": 3904
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0691620111465454,
      "learning_rate": 3.851173151492649e-06,
      "loss": 0.5513,
      "step": 3905
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1635345220565796,
      "learning_rate": 3.848805668975791e-06,
      "loss": 0.7721,
      "step": 3906
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.145581841468811,
      "learning_rate": 3.846438458971672e-06,
      "loss": 0.4571,
      "step": 3907
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1228947639465332,
      "learning_rate": 3.844071522040659e-06,
      "loss": 0.4706,
      "step": 3908
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1469179391860962,
      "learning_rate": 3.84170485874306e-06,
      "loss": 0.407,
      "step": 3909
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9643670320510864,
      "learning_rate": 3.8393384696391155e-06,
      "loss": 0.4423,
      "step": 3910
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1699268817901611,
      "learning_rate": 3.836972355289e-06,
      "loss": 0.411,
      "step": 3911
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0206961631774902,
      "learning_rate": 3.834606516252823e-06,
      "loss": 0.4754,
      "step": 3912
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5369856357574463,
      "learning_rate": 3.832240953090631e-06,
      "loss": 0.5135,
      "step": 3913
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1210469007492065,
      "learning_rate": 3.829875666362406e-06,
      "loss": 0.7312,
      "step": 3914
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1880615949630737,
      "learning_rate": 3.827510656628064e-06,
      "loss": 0.4514,
      "step": 3915
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3339414596557617,
      "learning_rate": 3.82514592444745e-06,
      "loss": 0.4805,
      "step": 3916
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3234200477600098,
      "learning_rate": 3.82278147038035e-06,
      "loss": 0.495,
      "step": 3917
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0900837182998657,
      "learning_rate": 3.820417294986482e-06,
      "loss": 0.4099,
      "step": 3918
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0673013925552368,
      "learning_rate": 3.818053398825496e-06,
      "loss": 0.4827,
      "step": 3919
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.134842038154602,
      "learning_rate": 3.815689782456979e-06,
      "loss": 0.4364,
      "step": 3920
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8573180437088013,
      "learning_rate": 3.8133264464404495e-06,
      "loss": 0.4905,
      "step": 3921
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2391581535339355,
      "learning_rate": 3.8109633913353627e-06,
      "loss": 0.4163,
      "step": 3922
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8302158117294312,
      "learning_rate": 3.8086006177011026e-06,
      "loss": 0.4869,
      "step": 3923
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1298677921295166,
      "learning_rate": 3.806238126096988e-06,
      "loss": 0.4663,
      "step": 3924
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3558400869369507,
      "learning_rate": 3.803875917082276e-06,
      "loss": 0.4615,
      "step": 3925
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0176714658737183,
      "learning_rate": 3.801513991216151e-06,
      "loss": 0.4395,
      "step": 3926
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2508628368377686,
      "learning_rate": 3.799152349057733e-06,
      "loss": 0.4443,
      "step": 3927
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.25247323513031,
      "learning_rate": 3.796790991166072e-06,
      "loss": 0.5073,
      "step": 3928
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.7466628551483154,
      "learning_rate": 3.7944299181001544e-06,
      "loss": 0.4469,
      "step": 3929
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1321390867233276,
      "learning_rate": 3.7920691304188973e-06,
      "loss": 0.4667,
      "step": 3930
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.225696086883545,
      "learning_rate": 3.7897086286811503e-06,
      "loss": 0.7255,
      "step": 3931
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.519807815551758,
      "learning_rate": 3.787348413445695e-06,
      "loss": 0.4617,
      "step": 3932
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2078956365585327,
      "learning_rate": 3.784988485271247e-06,
      "loss": 0.4631,
      "step": 3933
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.545699119567871,
      "learning_rate": 3.782628844716451e-06,
      "loss": 0.5563,
      "step": 3934
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0691031217575073,
      "learning_rate": 3.780269492339888e-06,
      "loss": 0.4005,
      "step": 3935
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4869250059127808,
      "learning_rate": 3.777910428700062e-06,
      "loss": 0.4075,
      "step": 3936
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.349518895149231,
      "learning_rate": 3.7755516543554223e-06,
      "loss": 0.506,
      "step": 3937
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1191331148147583,
      "learning_rate": 3.7731931698643377e-06,
      "loss": 0.4595,
      "step": 3938
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3369230031967163,
      "learning_rate": 3.7708349757851147e-06,
      "loss": 0.4915,
      "step": 3939
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3211374282836914,
      "learning_rate": 3.768477072675988e-06,
      "loss": 0.5124,
      "step": 3940
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2734076976776123,
      "learning_rate": 3.7661194610951235e-06,
      "loss": 0.4839,
      "step": 3941
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1572093963623047,
      "learning_rate": 3.763762141600621e-06,
      "loss": 0.486,
      "step": 3942
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.312811255455017,
      "learning_rate": 3.7614051147505083e-06,
      "loss": 0.7251,
      "step": 3943
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.299228549003601,
      "learning_rate": 3.759048381102745e-06,
      "loss": 0.4622,
      "step": 3944
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1870803833007812,
      "learning_rate": 3.756691941215222e-06,
      "loss": 0.4212,
      "step": 3945
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9606258869171143,
      "learning_rate": 3.7543357956457586e-06,
      "loss": 0.4465,
      "step": 3946
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9854576587677002,
      "learning_rate": 3.751979944952106e-06,
      "loss": 0.4876,
      "step": 3947
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2212175130844116,
      "learning_rate": 3.749624389691947e-06,
      "loss": 0.778,
      "step": 3948
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.004197120666504,
      "learning_rate": 3.7472691304228907e-06,
      "loss": 0.425,
      "step": 3949
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.737808346748352,
      "learning_rate": 3.7449141677024793e-06,
      "loss": 0.4783,
      "step": 3950
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.360809087753296,
      "learning_rate": 3.7425595020881844e-06,
      "loss": 0.4284,
      "step": 3951
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1733416318893433,
      "learning_rate": 3.7402051341374044e-06,
      "loss": 0.4074,
      "step": 3952
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.709716796875,
      "learning_rate": 3.7378510644074717e-06,
      "loss": 0.4525,
      "step": 3953
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9501498937606812,
      "learning_rate": 3.7354972934556435e-06,
      "loss": 0.5164,
      "step": 3954
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.632474660873413,
      "learning_rate": 3.7331438218391097e-06,
      "loss": 0.4642,
      "step": 3955
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3110042810440063,
      "learning_rate": 3.730790650114988e-06,
      "loss": 0.4576,
      "step": 3956
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1004769802093506,
      "learning_rate": 3.728437778840327e-06,
      "loss": 0.7123,
      "step": 3957
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1013414859771729,
      "learning_rate": 3.7260852085720987e-06,
      "loss": 0.4368,
      "step": 3958
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0740658044815063,
      "learning_rate": 3.7237329398672116e-06,
      "loss": 0.4086,
      "step": 3959
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1051584482192993,
      "learning_rate": 3.7213809732824973e-06,
      "loss": 0.4304,
      "step": 3960
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0849961042404175,
      "learning_rate": 3.7190293093747174e-06,
      "loss": 0.4227,
      "step": 3961
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.236849308013916,
      "learning_rate": 3.7166779487005623e-06,
      "loss": 0.4449,
      "step": 3962
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.8756790161132812,
      "learning_rate": 3.71432689181665e-06,
      "loss": 0.4957,
      "step": 3963
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.304988980293274,
      "learning_rate": 3.7119761392795265e-06,
      "loss": 0.4607,
      "step": 3964
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0761666297912598,
      "learning_rate": 3.7096256916456663e-06,
      "loss": 0.4734,
      "step": 3965
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.2683935165405273,
      "learning_rate": 3.7072755494714714e-06,
      "loss": 0.5242,
      "step": 3966
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4736288785934448,
      "learning_rate": 3.704925713313271e-06,
      "loss": 0.7117,
      "step": 3967
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2646723985671997,
      "learning_rate": 3.7025761837273235e-06,
      "loss": 0.461,
      "step": 3968
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4423801898956299,
      "learning_rate": 3.7002269612698118e-06,
      "loss": 0.4541,
      "step": 3969
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4017786979675293,
      "learning_rate": 3.6978780464968504e-06,
      "loss": 0.4233,
      "step": 3970
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.166804313659668,
      "learning_rate": 3.695529439964478e-06,
      "loss": 0.4371,
      "step": 3971
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0328328609466553,
      "learning_rate": 3.6931811422286596e-06,
      "loss": 0.4264,
      "step": 3972
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.296028733253479,
      "learning_rate": 3.690833153845289e-06,
      "loss": 0.4722,
      "step": 3973
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1666710376739502,
      "learning_rate": 3.688485475370187e-06,
      "loss": 0.4601,
      "step": 3974
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3510363101959229,
      "learning_rate": 3.6861381073590984e-06,
      "loss": 0.4783,
      "step": 3975
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5000532865524292,
      "learning_rate": 3.683791050367699e-06,
      "loss": 0.5054,
      "step": 3976
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2680625915527344,
      "learning_rate": 3.6814443049515846e-06,
      "loss": 0.4192,
      "step": 3977
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1409779787063599,
      "learning_rate": 3.679097871666283e-06,
      "loss": 0.4602,
      "step": 3978
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1967496871948242,
      "learning_rate": 3.6767517510672456e-06,
      "loss": 0.4923,
      "step": 3979
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2311122417449951,
      "learning_rate": 3.67440594370985e-06,
      "loss": 0.4825,
      "step": 3980
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0687141418457031,
      "learning_rate": 3.672060450149401e-06,
      "loss": 0.4539,
      "step": 3981
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.12493896484375,
      "learning_rate": 3.669715270941126e-06,
      "loss": 0.4983,
      "step": 3982
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1093511581420898,
      "learning_rate": 3.667370406640182e-06,
      "loss": 0.382,
      "step": 3983
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.083323359489441,
      "learning_rate": 3.6650258578016474e-06,
      "loss": 0.4738,
      "step": 3984
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7840447425842285,
      "learning_rate": 3.6626816249805295e-06,
      "loss": 0.418,
      "step": 3985
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3834115266799927,
      "learning_rate": 3.660337708731758e-06,
      "loss": 0.4798,
      "step": 3986
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2147191762924194,
      "learning_rate": 3.6579941096101908e-06,
      "loss": 0.5026,
      "step": 3987
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9803375601768494,
      "learning_rate": 3.655650828170607e-06,
      "loss": 0.4228,
      "step": 3988
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5858330726623535,
      "learning_rate": 3.6533078649677122e-06,
      "loss": 0.4843,
      "step": 3989
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8761742115020752,
      "learning_rate": 3.6509652205561376e-06,
      "loss": 0.4234,
      "step": 3990
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2009572982788086,
      "learning_rate": 3.6486228954904353e-06,
      "loss": 0.5178,
      "step": 3991
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4654699563980103,
      "learning_rate": 3.646280890325089e-06,
      "loss": 0.4355,
      "step": 3992
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1998015642166138,
      "learning_rate": 3.643939205614502e-06,
      "loss": 0.4353,
      "step": 3993
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6057642698287964,
      "learning_rate": 3.6415978419129983e-06,
      "loss": 0.4892,
      "step": 3994
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4349372386932373,
      "learning_rate": 3.639256799774832e-06,
      "loss": 0.446,
      "step": 3995
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5914047956466675,
      "learning_rate": 3.636916079754178e-06,
      "loss": 0.4466,
      "step": 3996
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.110493779182434,
      "learning_rate": 3.634575682405136e-06,
      "loss": 0.4916,
      "step": 3997
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3468079566955566,
      "learning_rate": 3.632235608281729e-06,
      "loss": 0.3892,
      "step": 3998
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.017630100250244,
      "learning_rate": 3.6298958579379034e-06,
      "loss": 0.4598,
      "step": 3999
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5705468654632568,
      "learning_rate": 3.62755643192753e-06,
      "loss": 0.455,
      "step": 4000
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3939602375030518,
      "learning_rate": 3.6252173308043997e-06,
      "loss": 0.4819,
      "step": 4001
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2258628606796265,
      "learning_rate": 3.622878555122228e-06,
      "loss": 0.5217,
      "step": 4002
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.440004825592041,
      "learning_rate": 3.6205401054346567e-06,
      "loss": 0.4849,
      "step": 4003
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0570286512374878,
      "learning_rate": 3.6182019822952474e-06,
      "loss": 0.5494,
      "step": 4004
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1730823516845703,
      "learning_rate": 3.615864186257485e-06,
      "loss": 0.4359,
      "step": 4005
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.112635374069214,
      "learning_rate": 3.6135267178747746e-06,
      "loss": 0.4241,
      "step": 4006
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3421692848205566,
      "learning_rate": 3.6111895777004463e-06,
      "loss": 0.5215,
      "step": 4007
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.011739730834961,
      "learning_rate": 3.6088527662877526e-06,
      "loss": 0.4882,
      "step": 4008
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.207665205001831,
      "learning_rate": 3.6065162841898683e-06,
      "loss": 0.4905,
      "step": 4009
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3148126602172852,
      "learning_rate": 3.604180131959888e-06,
      "loss": 0.5213,
      "step": 4010
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0288413763046265,
      "learning_rate": 3.6018443101508303e-06,
      "loss": 0.5287,
      "step": 4011
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.119321584701538,
      "learning_rate": 3.599508819315636e-06,
      "loss": 0.4886,
      "step": 4012
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.340699553489685,
      "learning_rate": 3.597173660007165e-06,
      "loss": 0.5352,
      "step": 4013
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1643693447113037,
      "learning_rate": 3.5948388327782015e-06,
      "loss": 0.3989,
      "step": 4014
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1959782838821411,
      "learning_rate": 3.5925043381814495e-06,
      "loss": 0.447,
      "step": 4015
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1209437847137451,
      "learning_rate": 3.5901701767695347e-06,
      "loss": 0.4323,
      "step": 4016
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0702043771743774,
      "learning_rate": 3.587836349095005e-06,
      "loss": 0.4273,
      "step": 4017
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3676462173461914,
      "learning_rate": 3.5855028557103253e-06,
      "loss": 0.5076,
      "step": 4018
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.258000135421753,
      "learning_rate": 3.583169697167887e-06,
      "loss": 0.4869,
      "step": 4019
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0816348791122437,
      "learning_rate": 3.580836874019998e-06,
      "loss": 0.4559,
      "step": 4020
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3267830610275269,
      "learning_rate": 3.5785043868188883e-06,
      "loss": 0.5371,
      "step": 4021
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2225703001022339,
      "learning_rate": 3.5761722361167095e-06,
      "loss": 0.4248,
      "step": 4022
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.200634002685547,
      "learning_rate": 3.573840422465531e-06,
      "loss": 0.4693,
      "step": 4023
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3368738889694214,
      "learning_rate": 3.5715089464173446e-06,
      "loss": 0.4523,
      "step": 4024
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3204604387283325,
      "learning_rate": 3.56917780852406e-06,
      "loss": 0.5188,
      "step": 4025
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1391935348510742,
      "learning_rate": 3.5668470093375113e-06,
      "loss": 0.505,
      "step": 4026
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2815417051315308,
      "learning_rate": 3.564516549409447e-06,
      "loss": 0.4898,
      "step": 4027
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1026663780212402,
      "learning_rate": 3.5621864292915383e-06,
      "loss": 0.4748,
      "step": 4028
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2680740356445312,
      "learning_rate": 3.559856649535376e-06,
      "loss": 0.4825,
      "step": 4029
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2989457845687866,
      "learning_rate": 3.5575272106924685e-06,
      "loss": 0.465,
      "step": 4030
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3453840017318726,
      "learning_rate": 3.5551981133142442e-06,
      "loss": 0.4392,
      "step": 4031
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5861550569534302,
      "learning_rate": 3.5528693579520523e-06,
      "loss": 0.5072,
      "step": 4032
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8562605381011963,
      "learning_rate": 3.5505409451571592e-06,
      "loss": 0.5309,
      "step": 4033
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1198326349258423,
      "learning_rate": 3.5482128754807517e-06,
      "loss": 0.4464,
      "step": 4034
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4091798067092896,
      "learning_rate": 3.545885149473933e-06,
      "loss": 0.4662,
      "step": 4035
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0370303392410278,
      "learning_rate": 3.5435577676877277e-06,
      "loss": 0.435,
      "step": 4036
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2140765190124512,
      "learning_rate": 3.5412307306730777e-06,
      "loss": 0.4244,
      "step": 4037
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1106606721878052,
      "learning_rate": 3.5389040389808427e-06,
      "loss": 0.4588,
      "step": 4038
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.249316930770874,
      "learning_rate": 3.536577693161801e-06,
      "loss": 0.4983,
      "step": 4039
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3299031257629395,
      "learning_rate": 3.5342516937666514e-06,
      "loss": 0.4388,
      "step": 4040
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0851243734359741,
      "learning_rate": 3.5319260413460053e-06,
      "loss": 0.4273,
      "step": 4041
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.19828462600708,
      "learning_rate": 3.5296007364503994e-06,
      "loss": 0.4994,
      "step": 4042
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2936116456985474,
      "learning_rate": 3.5272757796302793e-06,
      "loss": 0.4682,
      "step": 4043
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1976968050003052,
      "learning_rate": 3.524951171436015e-06,
      "loss": 0.422,
      "step": 4044
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1774492263793945,
      "learning_rate": 3.5226269124178923e-06,
      "loss": 0.4467,
      "step": 4045
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2422105073928833,
      "learning_rate": 3.520303003126113e-06,
      "loss": 0.505,
      "step": 4046
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1155712604522705,
      "learning_rate": 3.517979444110796e-06,
      "loss": 0.4436,
      "step": 4047
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.41659414768219,
      "learning_rate": 3.5156562359219808e-06,
      "loss": 0.5797,
      "step": 4048
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.185991883277893,
      "learning_rate": 3.5133333791096204e-06,
      "loss": 0.4579,
      "step": 4049
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9797555208206177,
      "learning_rate": 3.5110108742235847e-06,
      "loss": 0.4992,
      "step": 4050
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2897475957870483,
      "learning_rate": 3.508688721813661e-06,
      "loss": 0.7141,
      "step": 4051
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1873302459716797,
      "learning_rate": 3.506366922429555e-06,
      "loss": 0.4936,
      "step": 4052
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0960514545440674,
      "learning_rate": 3.5040454766208854e-06,
      "loss": 0.4682,
      "step": 4053
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1755198240280151,
      "learning_rate": 3.5017243849371914e-06,
      "loss": 0.4452,
      "step": 4054
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3498367071151733,
      "learning_rate": 3.4994036479279226e-06,
      "loss": 0.4808,
      "step": 4055
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5077916383743286,
      "learning_rate": 3.4970832661424485e-06,
      "loss": 0.4669,
      "step": 4056
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.174849033355713,
      "learning_rate": 3.4947632401300556e-06,
      "loss": 0.4445,
      "step": 4057
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.237277865409851,
      "learning_rate": 3.4924435704399414e-06,
      "loss": 0.5153,
      "step": 4058
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.467239499092102,
      "learning_rate": 3.4901242576212276e-06,
      "loss": 0.4895,
      "step": 4059
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.235283613204956,
      "learning_rate": 3.487805302222942e-06,
      "loss": 0.5401,
      "step": 4060
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4655460119247437,
      "learning_rate": 3.4854867047940324e-06,
      "loss": 0.4274,
      "step": 4061
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0092813968658447,
      "learning_rate": 3.4831684658833607e-06,
      "loss": 0.3974,
      "step": 4062
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2343941926956177,
      "learning_rate": 3.480850586039707e-06,
      "loss": 0.4887,
      "step": 4063
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1221023797988892,
      "learning_rate": 3.4785330658117606e-06,
      "loss": 0.3718,
      "step": 4064
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.261900782585144,
      "learning_rate": 3.4762159057481315e-06,
      "loss": 0.4663,
      "step": 4065
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.394280433654785,
      "learning_rate": 3.473899106397341e-06,
      "loss": 0.4996,
      "step": 4066
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.505173921585083,
      "learning_rate": 3.471582668307825e-06,
      "loss": 0.5175,
      "step": 4067
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1723802089691162,
      "learning_rate": 3.4692665920279355e-06,
      "loss": 0.3999,
      "step": 4068
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2569303512573242,
      "learning_rate": 3.466950878105937e-06,
      "loss": 0.5031,
      "step": 4069
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2952333688735962,
      "learning_rate": 3.4646355270900117e-06,
      "loss": 0.5355,
      "step": 4070
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.315492630004883,
      "learning_rate": 3.4623205395282533e-06,
      "loss": 0.5146,
      "step": 4071
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6158145666122437,
      "learning_rate": 3.4600059159686684e-06,
      "loss": 0.4301,
      "step": 4072
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9980474710464478,
      "learning_rate": 3.457691656959179e-06,
      "loss": 0.4154,
      "step": 4073
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6241710186004639,
      "learning_rate": 3.4553777630476205e-06,
      "loss": 0.4739,
      "step": 4074
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5471996068954468,
      "learning_rate": 3.4530642347817416e-06,
      "loss": 0.4183,
      "step": 4075
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3986074924468994,
      "learning_rate": 3.4507510727092055e-06,
      "loss": 0.4171,
      "step": 4076
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3822919130325317,
      "learning_rate": 3.4484382773775884e-06,
      "loss": 0.4569,
      "step": 4077
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5147806406021118,
      "learning_rate": 3.4461258493343773e-06,
      "loss": 0.4768,
      "step": 4078
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5737977027893066,
      "learning_rate": 3.4438137891269773e-06,
      "loss": 0.5433,
      "step": 4079
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1959490776062012,
      "learning_rate": 3.4415020973026985e-06,
      "loss": 0.4798,
      "step": 4080
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.392401933670044,
      "learning_rate": 3.4391907744087737e-06,
      "loss": 0.451,
      "step": 4081
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0884146690368652,
      "learning_rate": 3.4368798209923416e-06,
      "loss": 0.4562,
      "step": 4082
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3285495042800903,
      "learning_rate": 3.434569237600455e-06,
      "loss": 0.56,
      "step": 4083
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.96657395362854,
      "learning_rate": 3.4322590247800786e-06,
      "loss": 0.4776,
      "step": 4084
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0126820802688599,
      "learning_rate": 3.4299491830780907e-06,
      "loss": 0.4941,
      "step": 4085
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1340967416763306,
      "learning_rate": 3.42763971304128e-06,
      "loss": 0.4425,
      "step": 4086
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1141458749771118,
      "learning_rate": 3.4253306152163503e-06,
      "loss": 0.4882,
      "step": 4087
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1785550117492676,
      "learning_rate": 3.423021890149913e-06,
      "loss": 0.494,
      "step": 4088
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.572766900062561,
      "learning_rate": 3.420713538388495e-06,
      "loss": 0.4912,
      "step": 4089
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3400672674179077,
      "learning_rate": 3.4184055604785334e-06,
      "loss": 0.4084,
      "step": 4090
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1768999099731445,
      "learning_rate": 3.416097956966375e-06,
      "loss": 0.4645,
      "step": 4091
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4854992628097534,
      "learning_rate": 3.4137907283982817e-06,
      "loss": 0.4954,
      "step": 4092
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6043168306350708,
      "learning_rate": 3.411483875320425e-06,
      "loss": 0.4781,
      "step": 4093
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1906633377075195,
      "learning_rate": 3.4091773982788867e-06,
      "loss": 0.4606,
      "step": 4094
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1175004243850708,
      "learning_rate": 3.4068712978196593e-06,
      "loss": 0.4699,
      "step": 4095
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3746509552001953,
      "learning_rate": 3.4045655744886476e-06,
      "loss": 0.5266,
      "step": 4096
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.447356700897217,
      "learning_rate": 3.4022602288316654e-06,
      "loss": 0.5094,
      "step": 4097
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0234360694885254,
      "learning_rate": 3.39995526139444e-06,
      "loss": 0.3973,
      "step": 4098
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.36394464969635,
      "learning_rate": 3.3976506727226054e-06,
      "loss": 0.4449,
      "step": 4099
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4238781929016113,
      "learning_rate": 3.3953464633617085e-06,
      "loss": 0.4894,
      "step": 4100
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7979124784469604,
      "learning_rate": 3.3930426338572066e-06,
      "loss": 0.4982,
      "step": 4101
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4141911268234253,
      "learning_rate": 3.3907391847544656e-06,
      "loss": 0.4921,
      "step": 4102
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.692853331565857,
      "learning_rate": 3.388436116598761e-06,
      "loss": 0.4573,
      "step": 4103
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.226893424987793,
      "learning_rate": 3.386133429935281e-06,
      "loss": 0.4603,
      "step": 4104
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0247927904129028,
      "learning_rate": 3.3838311253091215e-06,
      "loss": 0.4904,
      "step": 4105
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0879428386688232,
      "learning_rate": 3.381529203265287e-06,
      "loss": 0.4257,
      "step": 4106
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2809725999832153,
      "learning_rate": 3.379227664348693e-06,
      "loss": 0.4695,
      "step": 4107
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1408029794692993,
      "learning_rate": 3.376926509104165e-06,
      "loss": 0.4025,
      "step": 4108
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0035349130630493,
      "learning_rate": 3.3746257380764335e-06,
      "loss": 0.4442,
      "step": 4109
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1468801498413086,
      "learning_rate": 3.372325351810144e-06,
      "loss": 0.5036,
      "step": 4110
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6976597309112549,
      "learning_rate": 3.370025350849846e-06,
      "loss": 0.4636,
      "step": 4111
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0990080833435059,
      "learning_rate": 3.3677257357400004e-06,
      "loss": 0.4717,
      "step": 4112
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.204171895980835,
      "learning_rate": 3.3654265070249766e-06,
      "loss": 0.4409,
      "step": 4113
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1632810831069946,
      "learning_rate": 3.363127665249051e-06,
      "loss": 0.4521,
      "step": 4114
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3532027006149292,
      "learning_rate": 3.36082921095641e-06,
      "loss": 0.4636,
      "step": 4115
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9888145327568054,
      "learning_rate": 3.3585311446911484e-06,
      "loss": 0.4396,
      "step": 4116
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.058267116546631,
      "learning_rate": 3.3562334669972684e-06,
      "loss": 0.4892,
      "step": 4117
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.19929838180542,
      "learning_rate": 3.353936178418679e-06,
      "loss": 0.5079,
      "step": 4118
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2177449464797974,
      "learning_rate": 3.3516392794992002e-06,
      "loss": 0.5053,
      "step": 4119
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5389232635498047,
      "learning_rate": 3.349342770782557e-06,
      "loss": 0.4749,
      "step": 4120
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0123933553695679,
      "learning_rate": 3.3470466528123822e-06,
      "loss": 0.4141,
      "step": 4121
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.221616268157959,
      "learning_rate": 3.3447509261322174e-06,
      "loss": 0.4036,
      "step": 4122
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.6743993759155273,
      "learning_rate": 3.3424555912855116e-06,
      "loss": 0.4674,
      "step": 4123
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.223986268043518,
      "learning_rate": 3.34016064881562e-06,
      "loss": 0.498,
      "step": 4124
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5775165557861328,
      "learning_rate": 3.3378660992658042e-06,
      "loss": 0.4583,
      "step": 4125
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5224918127059937,
      "learning_rate": 3.3355719431792367e-06,
      "loss": 0.461,
      "step": 4126
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.561614036560059,
      "learning_rate": 3.333278181098992e-06,
      "loss": 0.505,
      "step": 4127
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2010118961334229,
      "learning_rate": 3.3309848135680544e-06,
      "loss": 0.4196,
      "step": 4128
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.106626033782959,
      "learning_rate": 3.3286918411293143e-06,
      "loss": 0.4766,
      "step": 4129
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2585874795913696,
      "learning_rate": 3.326399264325567e-06,
      "loss": 0.4442,
      "step": 4130
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2040386199951172,
      "learning_rate": 3.324107083699516e-06,
      "loss": 0.4654,
      "step": 4131
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4107829332351685,
      "learning_rate": 3.321815299793771e-06,
      "loss": 0.4836,
      "step": 4132
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5626001358032227,
      "learning_rate": 3.319523913150845e-06,
      "loss": 0.4656,
      "step": 4133
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1086653470993042,
      "learning_rate": 3.3172329243131606e-06,
      "loss": 0.7321,
      "step": 4134
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0947980880737305,
      "learning_rate": 3.3149423338230437e-06,
      "loss": 0.5197,
      "step": 4135
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3430309295654297,
      "learning_rate": 3.3126521422227264e-06,
      "loss": 0.4573,
      "step": 4136
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.007129192352295,
      "learning_rate": 3.3103623500543504e-06,
      "loss": 0.3924,
      "step": 4137
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.230101466178894,
      "learning_rate": 3.308072957859956e-06,
      "loss": 0.4653,
      "step": 4138
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3543610572814941,
      "learning_rate": 3.305783966181493e-06,
      "loss": 0.4979,
      "step": 4139
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2025514841079712,
      "learning_rate": 3.3034953755608155e-06,
      "loss": 0.7207,
      "step": 4140
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0665555000305176,
      "learning_rate": 3.3012071865396835e-06,
      "loss": 0.4741,
      "step": 4141
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2860620021820068,
      "learning_rate": 3.2989193996597604e-06,
      "loss": 0.4582,
      "step": 4142
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9481910467147827,
      "learning_rate": 3.296632015462614e-06,
      "loss": 0.4645,
      "step": 4143
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1811827421188354,
      "learning_rate": 3.2943450344897206e-06,
      "loss": 0.474,
      "step": 4144
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2908374071121216,
      "learning_rate": 3.2920584572824556e-06,
      "loss": 0.4744,
      "step": 4145
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.1159539222717285,
      "learning_rate": 3.289772284382102e-06,
      "loss": 0.4447,
      "step": 4146
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1471368074417114,
      "learning_rate": 3.2874865163298466e-06,
      "loss": 0.455,
      "step": 4147
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.2193398475646973,
      "learning_rate": 3.285201153666783e-06,
      "loss": 0.3939,
      "step": 4148
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2730371952056885,
      "learning_rate": 3.282916196933904e-06,
      "loss": 0.732,
      "step": 4149
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3616809844970703,
      "learning_rate": 3.280631646672109e-06,
      "loss": 0.4817,
      "step": 4150
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2298057079315186,
      "learning_rate": 3.2783475034222007e-06,
      "loss": 0.4159,
      "step": 4151
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0002925395965576,
      "learning_rate": 3.276063767724885e-06,
      "loss": 0.4621,
      "step": 4152
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.035576820373535,
      "learning_rate": 3.273780440120773e-06,
      "loss": 0.4125,
      "step": 4153
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1105072498321533,
      "learning_rate": 3.2714975211503765e-06,
      "loss": 0.7242,
      "step": 4154
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3407618999481201,
      "learning_rate": 3.269215011354113e-06,
      "loss": 0.4883,
      "step": 4155
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7654086351394653,
      "learning_rate": 3.2669329112723025e-06,
      "loss": 0.4868,
      "step": 4156
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1390154361724854,
      "learning_rate": 3.264651221445168e-06,
      "loss": 0.4299,
      "step": 4157
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2657530307769775,
      "learning_rate": 3.262369942412832e-06,
      "loss": 0.4701,
      "step": 4158
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.923278331756592,
      "learning_rate": 3.2600890747153264e-06,
      "loss": 0.4631,
      "step": 4159
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1033730506896973,
      "learning_rate": 3.257808618892581e-06,
      "loss": 0.4365,
      "step": 4160
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3348567485809326,
      "learning_rate": 3.2555285754844313e-06,
      "loss": 0.4843,
      "step": 4161
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5389865636825562,
      "learning_rate": 3.2532489450306093e-06,
      "loss": 0.4475,
      "step": 4162
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3229706287384033,
      "learning_rate": 3.2509697280707554e-06,
      "loss": 0.3996,
      "step": 4163
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1268466711044312,
      "learning_rate": 3.2486909251444097e-06,
      "loss": 0.4257,
      "step": 4164
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.408193826675415,
      "learning_rate": 3.2464125367910137e-06,
      "loss": 0.458,
      "step": 4165
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2430939674377441,
      "learning_rate": 3.2441345635499125e-06,
      "loss": 0.437,
      "step": 4166
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.121511459350586,
      "learning_rate": 3.2418570059603516e-06,
      "loss": 0.7187,
      "step": 4167
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0278691053390503,
      "learning_rate": 3.2395798645614775e-06,
      "loss": 0.6769,
      "step": 4168
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1349406242370605,
      "learning_rate": 3.2373031398923394e-06,
      "loss": 0.4887,
      "step": 4169
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.649074673652649,
      "learning_rate": 3.235026832491889e-06,
      "loss": 0.5025,
      "step": 4170
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0113850831985474,
      "learning_rate": 3.2327509428989755e-06,
      "loss": 0.3887,
      "step": 4171
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.208206057548523,
      "learning_rate": 3.230475471652354e-06,
      "loss": 0.7314,
      "step": 4172
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3545582294464111,
      "learning_rate": 3.2282004192906774e-06,
      "loss": 0.4391,
      "step": 4173
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.197930097579956,
      "learning_rate": 3.2259257863524983e-06,
      "loss": 0.5214,
      "step": 4174
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.381136417388916,
      "learning_rate": 3.223651573376273e-06,
      "loss": 0.4587,
      "step": 4175
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1372946500778198,
      "learning_rate": 3.2213777809003567e-06,
      "loss": 0.4701,
      "step": 4176
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1420704126358032,
      "learning_rate": 3.2191044094630064e-06,
      "loss": 0.4,
      "step": 4177
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0223647356033325,
      "learning_rate": 3.2168314596023774e-06,
      "loss": 0.3941,
      "step": 4178
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3392431735992432,
      "learning_rate": 3.214558931856528e-06,
      "loss": 0.4091,
      "step": 4179
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2686574459075928,
      "learning_rate": 3.212286826763412e-06,
      "loss": 0.4731,
      "step": 4180
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.7402193546295166,
      "learning_rate": 3.21001514486089e-06,
      "loss": 0.4438,
      "step": 4181
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1724262237548828,
      "learning_rate": 3.2077438866867163e-06,
      "loss": 0.4871,
      "step": 4182
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.281681776046753,
      "learning_rate": 3.205473052778547e-06,
      "loss": 0.4969,
      "step": 4183
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.075587511062622,
      "learning_rate": 3.2032026436739393e-06,
      "loss": 0.5124,
      "step": 4184
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1959187984466553,
      "learning_rate": 3.2009326599103467e-06,
      "loss": 0.4636,
      "step": 4185
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5312252044677734,
      "learning_rate": 3.1986631020251257e-06,
      "loss": 0.5089,
      "step": 4186
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.4839065074920654,
      "learning_rate": 3.1963939705555274e-06,
      "loss": 0.4512,
      "step": 4187
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4848909378051758,
      "learning_rate": 3.194125266038706e-06,
      "loss": 0.4671,
      "step": 4188
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.234470009803772,
      "learning_rate": 3.1918569890117137e-06,
      "loss": 0.3886,
      "step": 4189
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.277927279472351,
      "learning_rate": 3.1895891400114997e-06,
      "loss": 0.5071,
      "step": 4190
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1254165172576904,
      "learning_rate": 3.1873217195749144e-06,
      "loss": 0.4575,
      "step": 4191
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.79015052318573,
      "learning_rate": 3.185054728238704e-06,
      "loss": 0.4401,
      "step": 4192
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2739313840866089,
      "learning_rate": 3.1827881665395167e-06,
      "loss": 0.4636,
      "step": 4193
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3658758401870728,
      "learning_rate": 3.180522035013896e-06,
      "loss": 0.4334,
      "step": 4194
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.466949462890625,
      "learning_rate": 3.178256334198284e-06,
      "loss": 0.4271,
      "step": 4195
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3730446100234985,
      "learning_rate": 3.1759910646290227e-06,
      "loss": 0.485,
      "step": 4196
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.68057119846344,
      "learning_rate": 3.1737262268423495e-06,
      "loss": 0.4891,
      "step": 4197
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1726239919662476,
      "learning_rate": 3.1714618213744025e-06,
      "loss": 0.7259,
      "step": 4198
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4657522439956665,
      "learning_rate": 3.1691978487612127e-06,
      "loss": 0.5331,
      "step": 4199
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1113135814666748,
      "learning_rate": 3.1669343095387138e-06,
      "loss": 0.5092,
      "step": 4200
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4178999662399292,
      "learning_rate": 3.164671204242734e-06,
      "loss": 0.4217,
      "step": 4201
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.023424744606018,
      "learning_rate": 3.162408533409e-06,
      "loss": 0.7072,
      "step": 4202
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2182066440582275,
      "learning_rate": 3.1601462975731336e-06,
      "loss": 0.491,
      "step": 4203
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1490273475646973,
      "learning_rate": 3.157884497270658e-06,
      "loss": 0.4728,
      "step": 4204
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0976747274398804,
      "learning_rate": 3.155623133036989e-06,
      "loss": 0.6967,
      "step": 4205
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3245422840118408,
      "learning_rate": 3.153362205407441e-06,
      "loss": 0.4385,
      "step": 4206
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.401752233505249,
      "learning_rate": 3.1511017149172247e-06,
      "loss": 0.4519,
      "step": 4207
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1377710103988647,
      "learning_rate": 3.1488416621014474e-06,
      "loss": 0.7072,
      "step": 4208
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4046118259429932,
      "learning_rate": 3.146582047495112e-06,
      "loss": 0.4384,
      "step": 4209
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3487035036087036,
      "learning_rate": 3.144322871633121e-06,
      "loss": 0.3924,
      "step": 4210
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.357029676437378,
      "learning_rate": 3.1420641350502667e-06,
      "loss": 0.4122,
      "step": 4211
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.135287046432495,
      "learning_rate": 3.1398058382812435e-06,
      "loss": 0.5042,
      "step": 4212
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.043498158454895,
      "learning_rate": 3.1375479818606387e-06,
      "loss": 0.4515,
      "step": 4213
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.39207124710083,
      "learning_rate": 3.135290566322934e-06,
      "loss": 0.4788,
      "step": 4214
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1720491647720337,
      "learning_rate": 3.1330335922025148e-06,
      "loss": 0.401,
      "step": 4215
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7058765888214111,
      "learning_rate": 3.1307770600336496e-06,
      "loss": 0.4826,
      "step": 4216
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3872730731964111,
      "learning_rate": 3.1285209703505126e-06,
      "loss": 0.5378,
      "step": 4217
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.184346318244934,
      "learning_rate": 3.126265323687167e-06,
      "loss": 0.4502,
      "step": 4218
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.341003179550171,
      "learning_rate": 3.1240101205775748e-06,
      "loss": 0.44,
      "step": 4219
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0986262559890747,
      "learning_rate": 3.1217553615555906e-06,
      "loss": 0.4531,
      "step": 4220
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.6734561920166016,
      "learning_rate": 3.119501047154966e-06,
      "loss": 0.5056,
      "step": 4221
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4704256057739258,
      "learning_rate": 3.117247177909346e-06,
      "loss": 0.4976,
      "step": 4222
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8665271997451782,
      "learning_rate": 3.1149937543522702e-06,
      "loss": 0.4822,
      "step": 4223
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.455519437789917,
      "learning_rate": 3.1127407770171726e-06,
      "loss": 0.4196,
      "step": 4224
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0243233442306519,
      "learning_rate": 3.11048824643738e-06,
      "loss": 0.4739,
      "step": 4225
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6871980428695679,
      "learning_rate": 3.10823616314612e-06,
      "loss": 0.4997,
      "step": 4226
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3386026620864868,
      "learning_rate": 3.1059845276765077e-06,
      "loss": 0.4417,
      "step": 4227
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3005279302597046,
      "learning_rate": 3.1037333405615535e-06,
      "loss": 0.5429,
      "step": 4228
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0612643957138062,
      "learning_rate": 3.101482602334162e-06,
      "loss": 0.5457,
      "step": 4229
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3234763145446777,
      "learning_rate": 3.099232313527133e-06,
      "loss": 0.4488,
      "step": 4230
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3210493326187134,
      "learning_rate": 3.096982474673158e-06,
      "loss": 0.4492,
      "step": 4231
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2100218534469604,
      "learning_rate": 3.094733086304824e-06,
      "loss": 0.4653,
      "step": 4232
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5024646520614624,
      "learning_rate": 3.092484148954609e-06,
      "loss": 0.4636,
      "step": 4233
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4540656805038452,
      "learning_rate": 3.090235663154885e-06,
      "loss": 0.4408,
      "step": 4234
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.172984004020691,
      "learning_rate": 3.0879876294379197e-06,
      "loss": 0.4629,
      "step": 4235
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.139290690422058,
      "learning_rate": 3.0857400483358676e-06,
      "loss": 0.4675,
      "step": 4236
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.222582221031189,
      "learning_rate": 3.083492920380784e-06,
      "loss": 0.6592,
      "step": 4237
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7302050590515137,
      "learning_rate": 3.0812462461046116e-06,
      "loss": 0.4833,
      "step": 4238
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2995322942733765,
      "learning_rate": 3.0790000260391884e-06,
      "loss": 0.5284,
      "step": 4239
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2683719396591187,
      "learning_rate": 3.0767542607162405e-06,
      "loss": 0.4565,
      "step": 4240
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.25870680809021,
      "learning_rate": 3.074508950667391e-06,
      "loss": 0.5263,
      "step": 4241
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.87152099609375,
      "learning_rate": 3.0722640964241534e-06,
      "loss": 0.4718,
      "step": 4242
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5497099161148071,
      "learning_rate": 3.0700196985179343e-06,
      "loss": 0.4336,
      "step": 4243
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1340137720108032,
      "learning_rate": 3.067775757480031e-06,
      "loss": 0.4463,
      "step": 4244
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1072421073913574,
      "learning_rate": 3.0655322738416325e-06,
      "loss": 0.3546,
      "step": 4245
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1650868654251099,
      "learning_rate": 3.063289248133822e-06,
      "loss": 0.6788,
      "step": 4246
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5586978197097778,
      "learning_rate": 3.0610466808875693e-06,
      "loss": 0.4615,
      "step": 4247
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1743648052215576,
      "learning_rate": 3.058804572633742e-06,
      "loss": 0.5032,
      "step": 4248
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.520743489265442,
      "learning_rate": 3.0565629239030946e-06,
      "loss": 0.5023,
      "step": 4249
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.068416714668274,
      "learning_rate": 3.0543217352262746e-06,
      "loss": 0.4664,
      "step": 4250
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5580904483795166,
      "learning_rate": 3.052081007133819e-06,
      "loss": 0.4433,
      "step": 4251
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3232862949371338,
      "learning_rate": 3.0498407401561586e-06,
      "loss": 0.4388,
      "step": 4252
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.86197829246521,
      "learning_rate": 3.0476009348236117e-06,
      "loss": 0.4101,
      "step": 4253
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2948298454284668,
      "learning_rate": 3.0453615916663894e-06,
      "loss": 0.4308,
      "step": 4254
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2773562669754028,
      "learning_rate": 3.0431227112145933e-06,
      "loss": 0.4565,
      "step": 4255
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.795466661453247,
      "learning_rate": 3.040884293998215e-06,
      "loss": 0.4122,
      "step": 4256
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.144334077835083,
      "learning_rate": 3.0386463405471362e-06,
      "loss": 0.4536,
      "step": 4257
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3571081161499023,
      "learning_rate": 3.0364088513911283e-06,
      "loss": 0.4669,
      "step": 4258
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4785912036895752,
      "learning_rate": 3.0341718270598557e-06,
      "loss": 0.461,
      "step": 4259
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.231850266456604,
      "learning_rate": 3.0319352680828696e-06,
      "loss": 0.4539,
      "step": 4260
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4127272367477417,
      "learning_rate": 3.0296991749896133e-06,
      "loss": 0.4777,
      "step": 4261
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2510863542556763,
      "learning_rate": 3.027463548309416e-06,
      "loss": 0.4978,
      "step": 4262
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4076340198516846,
      "learning_rate": 3.0252283885715024e-06,
      "loss": 0.4721,
      "step": 4263
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.291025996208191,
      "learning_rate": 3.0229936963049824e-06,
      "loss": 0.4637,
      "step": 4264
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.2256155014038086,
      "learning_rate": 3.0207594720388546e-06,
      "loss": 0.4282,
      "step": 4265
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3109581470489502,
      "learning_rate": 3.018525716302009e-06,
      "loss": 0.4954,
      "step": 4266
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4703086614608765,
      "learning_rate": 3.0162924296232254e-06,
      "loss": 0.4948,
      "step": 4267
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2645330429077148,
      "learning_rate": 3.0140596125311707e-06,
      "loss": 0.4365,
      "step": 4268
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3917535543441772,
      "learning_rate": 3.011827265554401e-06,
      "loss": 0.4241,
      "step": 4269
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5185726881027222,
      "learning_rate": 3.0095953892213615e-06,
      "loss": 0.5064,
      "step": 4270
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1748316287994385,
      "learning_rate": 3.0073639840603863e-06,
      "loss": 0.4448,
      "step": 4271
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5313670635223389,
      "learning_rate": 3.0051330505996973e-06,
      "loss": 0.5323,
      "step": 4272
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7067928314208984,
      "learning_rate": 3.002902589367405e-06,
      "loss": 0.4486,
      "step": 4273
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2149091958999634,
      "learning_rate": 3.0006726008915077e-06,
      "loss": 0.5232,
      "step": 4274
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4726197719573975,
      "learning_rate": 2.9984430856998934e-06,
      "loss": 0.45,
      "step": 4275
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0027389526367188,
      "learning_rate": 2.9962140443203363e-06,
      "loss": 0.5101,
      "step": 4276
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5061918497085571,
      "learning_rate": 2.9939854772804977e-06,
      "loss": 0.4548,
      "step": 4277
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1325739622116089,
      "learning_rate": 2.991757385107929e-06,
      "loss": 0.4658,
      "step": 4278
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.089577078819275,
      "learning_rate": 2.989529768330068e-06,
      "loss": 0.6973,
      "step": 4279
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.576018214225769,
      "learning_rate": 2.98730262747424e-06,
      "loss": 0.4077,
      "step": 4280
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4859997034072876,
      "learning_rate": 2.985075963067656e-06,
      "loss": 0.4907,
      "step": 4281
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.424350619316101,
      "learning_rate": 2.9828497756374184e-06,
      "loss": 0.6109,
      "step": 4282
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7341022491455078,
      "learning_rate": 2.9806240657105144e-06,
      "loss": 0.5189,
      "step": 4283
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6717169284820557,
      "learning_rate": 2.9783988338138157e-06,
      "loss": 0.5188,
      "step": 4284
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4115805625915527,
      "learning_rate": 2.9761740804740856e-06,
      "loss": 0.4965,
      "step": 4285
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.495171070098877,
      "learning_rate": 2.9739498062179696e-06,
      "loss": 0.4227,
      "step": 4286
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.048019528388977,
      "learning_rate": 2.971726011572002e-06,
      "loss": 0.4314,
      "step": 4287
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.264007568359375,
      "learning_rate": 2.969502697062605e-06,
      "loss": 0.442,
      "step": 4288
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2750391960144043,
      "learning_rate": 2.967279863216084e-06,
      "loss": 0.4582,
      "step": 4289
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3466196060180664,
      "learning_rate": 2.9650575105586314e-06,
      "loss": 0.4348,
      "step": 4290
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.533465027809143,
      "learning_rate": 2.9628356396163283e-06,
      "loss": 0.4418,
      "step": 4291
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.593993902206421,
      "learning_rate": 2.9606142509151364e-06,
      "loss": 0.4996,
      "step": 4292
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0439900159835815,
      "learning_rate": 2.9583933449809128e-06,
      "loss": 0.3966,
      "step": 4293
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.7913713455200195,
      "learning_rate": 2.95617292233939e-06,
      "loss": 0.509,
      "step": 4294
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3553026914596558,
      "learning_rate": 2.95395298351619e-06,
      "loss": 0.4497,
      "step": 4295
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3052610158920288,
      "learning_rate": 2.9517335290368227e-06,
      "loss": 0.418,
      "step": 4296
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.314295768737793,
      "learning_rate": 2.94951455942668e-06,
      "loss": 0.5395,
      "step": 4297
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2944961786270142,
      "learning_rate": 2.9472960752110404e-06,
      "loss": 0.4064,
      "step": 4298
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3246731758117676,
      "learning_rate": 2.9450780769150675e-06,
      "loss": 0.4558,
      "step": 4299
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.133744239807129,
      "learning_rate": 2.94286056506381e-06,
      "loss": 0.7469,
      "step": 4300
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7805116176605225,
      "learning_rate": 2.940643540182202e-06,
      "loss": 0.4744,
      "step": 4301
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4731719493865967,
      "learning_rate": 2.9384270027950594e-06,
      "loss": 0.4366,
      "step": 4302
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1789754629135132,
      "learning_rate": 2.9362109534270844e-06,
      "loss": 0.4435,
      "step": 4303
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4314136505126953,
      "learning_rate": 2.9339953926028664e-06,
      "loss": 0.4279,
      "step": 4304
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5375174283981323,
      "learning_rate": 2.9317803208468777e-06,
      "loss": 0.4316,
      "step": 4305
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3275443315505981,
      "learning_rate": 2.929565738683471e-06,
      "loss": 0.5117,
      "step": 4306
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8359519243240356,
      "learning_rate": 2.9273516466368877e-06,
      "loss": 0.5198,
      "step": 4307
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.300997018814087,
      "learning_rate": 2.925138045231251e-06,
      "loss": 0.4122,
      "step": 4308
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.240480899810791,
      "learning_rate": 2.9229249349905686e-06,
      "loss": 0.6794,
      "step": 4309
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2396682500839233,
      "learning_rate": 2.9207123164387318e-06,
      "loss": 0.5284,
      "step": 4310
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.207517147064209,
      "learning_rate": 2.9185001900995168e-06,
      "loss": 0.4353,
      "step": 4311
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.3722267150878906,
      "learning_rate": 2.916288556496579e-06,
      "loss": 0.3813,
      "step": 4312
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1722626686096191,
      "learning_rate": 2.9140774161534633e-06,
      "loss": 0.4873,
      "step": 4313
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2547613382339478,
      "learning_rate": 2.911866769593592e-06,
      "loss": 0.4985,
      "step": 4314
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2361258268356323,
      "learning_rate": 2.909656617340275e-06,
      "loss": 0.7746,
      "step": 4315
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.632564544677734,
      "learning_rate": 2.9074469599167045e-06,
      "loss": 0.4564,
      "step": 4316
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.00584876537323,
      "learning_rate": 2.9052377978459512e-06,
      "loss": 0.4951,
      "step": 4317
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.253683090209961,
      "learning_rate": 2.9030291316509756e-06,
      "loss": 0.4652,
      "step": 4318
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3260748386383057,
      "learning_rate": 2.9008209618546134e-06,
      "loss": 0.4569,
      "step": 4319
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.213498830795288,
      "learning_rate": 2.89861328897959e-06,
      "loss": 0.511,
      "step": 4320
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.7926478385925293,
      "learning_rate": 2.8964061135485076e-06,
      "loss": 0.5242,
      "step": 4321
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3149771690368652,
      "learning_rate": 2.894199436083851e-06,
      "loss": 0.4416,
      "step": 4322
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2420238256454468,
      "learning_rate": 2.8919932571079925e-06,
      "loss": 0.4329,
      "step": 4323
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4246973991394043,
      "learning_rate": 2.8897875771431778e-06,
      "loss": 0.4303,
      "step": 4324
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2363532781600952,
      "learning_rate": 2.8875823967115426e-06,
      "loss": 0.4717,
      "step": 4325
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.057025671005249,
      "learning_rate": 2.8853777163351027e-06,
      "loss": 0.4351,
      "step": 4326
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4380855560302734,
      "learning_rate": 2.88317353653575e-06,
      "loss": 0.4628,
      "step": 4327
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.140413522720337,
      "learning_rate": 2.880969857835266e-06,
      "loss": 0.4899,
      "step": 4328
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.438979148864746,
      "learning_rate": 2.878766680755304e-06,
      "loss": 0.4275,
      "step": 4329
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2649264335632324,
      "learning_rate": 2.8765640058174096e-06,
      "loss": 0.4816,
      "step": 4330
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5692439079284668,
      "learning_rate": 2.874361833542999e-06,
      "loss": 0.5383,
      "step": 4331
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4346257448196411,
      "learning_rate": 2.8721601644533787e-06,
      "loss": 0.5724,
      "step": 4332
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2802366018295288,
      "learning_rate": 2.8699589990697274e-06,
      "loss": 0.4702,
      "step": 4333
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3037101030349731,
      "learning_rate": 2.8677583379131124e-06,
      "loss": 0.4887,
      "step": 4334
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1260584592819214,
      "learning_rate": 2.8655581815044775e-06,
      "loss": 0.4313,
      "step": 4335
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.073386549949646,
      "learning_rate": 2.8633585303646414e-06,
      "loss": 0.4749,
      "step": 4336
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1635862588882446,
      "learning_rate": 2.8611593850143197e-06,
      "loss": 0.4111,
      "step": 4337
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.261318325996399,
      "learning_rate": 2.8589607459740924e-06,
      "loss": 0.4718,
      "step": 4338
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1678438186645508,
      "learning_rate": 2.8567626137644232e-06,
      "loss": 0.459,
      "step": 4339
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0424017906188965,
      "learning_rate": 2.854564988905663e-06,
      "loss": 0.4306,
      "step": 4340
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2831569910049438,
      "learning_rate": 2.8523678719180325e-06,
      "loss": 0.3935,
      "step": 4341
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1731399297714233,
      "learning_rate": 2.8501712633216403e-06,
      "loss": 0.431,
      "step": 4342
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4413591623306274,
      "learning_rate": 2.8479751636364693e-06,
      "loss": 0.4305,
      "step": 4343
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.215792179107666,
      "learning_rate": 2.845779573382387e-06,
      "loss": 0.4838,
      "step": 4344
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3511093854904175,
      "learning_rate": 2.8435844930791325e-06,
      "loss": 0.4723,
      "step": 4345
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.659837245941162,
      "learning_rate": 2.841389923246335e-06,
      "loss": 0.4407,
      "step": 4346
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2580485343933105,
      "learning_rate": 2.839195864403491e-06,
      "loss": 0.4478,
      "step": 4347
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9443379044532776,
      "learning_rate": 2.8370023170699856e-06,
      "loss": 0.4094,
      "step": 4348
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2230851650238037,
      "learning_rate": 2.834809281765081e-06,
      "loss": 0.4839,
      "step": 4349
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1464214324951172,
      "learning_rate": 2.8326167590079123e-06,
      "loss": 0.5326,
      "step": 4350
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.032422423362732,
      "learning_rate": 2.830424749317498e-06,
      "loss": 0.4547,
      "step": 4351
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3530924320220947,
      "learning_rate": 2.8282332532127367e-06,
      "loss": 0.4551,
      "step": 4352
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9910420179367065,
      "learning_rate": 2.8260422712124e-06,
      "loss": 0.4766,
      "step": 4353
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2903687953948975,
      "learning_rate": 2.8238518038351448e-06,
      "loss": 0.4689,
      "step": 4354
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0930019617080688,
      "learning_rate": 2.821661851599498e-06,
      "loss": 0.4238,
      "step": 4355
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.381022334098816,
      "learning_rate": 2.819472415023874e-06,
      "loss": 0.4259,
      "step": 4356
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1900492906570435,
      "learning_rate": 2.8172834946265547e-06,
      "loss": 0.4343,
      "step": 4357
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0985747575759888,
      "learning_rate": 2.81509509092571e-06,
      "loss": 0.5179,
      "step": 4358
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2399247884750366,
      "learning_rate": 2.8129072044393785e-06,
      "loss": 0.4937,
      "step": 4359
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3127015829086304,
      "learning_rate": 2.8107198356854827e-06,
      "loss": 0.4705,
      "step": 4360
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3152960538864136,
      "learning_rate": 2.8085329851818215e-06,
      "loss": 0.4654,
      "step": 4361
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.9074029922485352,
      "learning_rate": 2.806346653446067e-06,
      "loss": 0.5318,
      "step": 4362
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6221473217010498,
      "learning_rate": 2.8041608409957753e-06,
      "loss": 0.4414,
      "step": 4363
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.778001070022583,
      "learning_rate": 2.8019755483483735e-06,
      "loss": 0.4792,
      "step": 4364
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.47916579246521,
      "learning_rate": 2.799790776021165e-06,
      "loss": 0.447,
      "step": 4365
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3043886423110962,
      "learning_rate": 2.7976065245313388e-06,
      "loss": 0.7583,
      "step": 4366
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3503074645996094,
      "learning_rate": 2.79542279439595e-06,
      "loss": 0.5175,
      "step": 4367
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0536876916885376,
      "learning_rate": 2.7932395861319385e-06,
      "loss": 0.4254,
      "step": 4368
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1929293870925903,
      "learning_rate": 2.7910569002561137e-06,
      "loss": 0.4767,
      "step": 4369
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6056745052337646,
      "learning_rate": 2.7888747372851666e-06,
      "loss": 0.4952,
      "step": 4370
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0783488750457764,
      "learning_rate": 2.7866930977356644e-06,
      "loss": 0.7099,
      "step": 4371
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1096841096878052,
      "learning_rate": 2.7845119821240453e-06,
      "loss": 0.4628,
      "step": 4372
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2432806491851807,
      "learning_rate": 2.7823313909666295e-06,
      "loss": 0.4492,
      "step": 4373
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0852199792861938,
      "learning_rate": 2.780151324779608e-06,
      "loss": 0.4577,
      "step": 4374
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2968734502792358,
      "learning_rate": 2.7779717840790532e-06,
      "loss": 0.4644,
      "step": 4375
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5442878007888794,
      "learning_rate": 2.7757927693809074e-06,
      "loss": 0.4789,
      "step": 4376
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4198272228240967,
      "learning_rate": 2.77361428120099e-06,
      "loss": 0.4783,
      "step": 4377
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4411635398864746,
      "learning_rate": 2.7714363200549986e-06,
      "loss": 0.4522,
      "step": 4378
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0620163679122925,
      "learning_rate": 2.769258886458501e-06,
      "loss": 0.4701,
      "step": 4379
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.29343581199646,
      "learning_rate": 2.767081980926948e-06,
      "loss": 0.5555,
      "step": 4380
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3196594715118408,
      "learning_rate": 2.764905603975655e-06,
      "loss": 0.5468,
      "step": 4381
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0906745195388794,
      "learning_rate": 2.7627297561198206e-06,
      "loss": 0.444,
      "step": 4382
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3168686628341675,
      "learning_rate": 2.7605544378745176e-06,
      "loss": 0.3778,
      "step": 4383
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3309441804885864,
      "learning_rate": 2.7583796497546855e-06,
      "loss": 0.4505,
      "step": 4384
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3119064569473267,
      "learning_rate": 2.75620539227515e-06,
      "loss": 0.4431,
      "step": 4385
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1982358694076538,
      "learning_rate": 2.7540316659506e-06,
      "loss": 0.5008,
      "step": 4386
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.266230821609497,
      "learning_rate": 2.7518584712956074e-06,
      "loss": 0.4433,
      "step": 4387
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3082969188690186,
      "learning_rate": 2.7496858088246144e-06,
      "loss": 0.4468,
      "step": 4388
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3101959228515625,
      "learning_rate": 2.7475136790519334e-06,
      "loss": 0.5162,
      "step": 4389
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.179983139038086,
      "learning_rate": 2.74534208249176e-06,
      "loss": 0.4733,
      "step": 4390
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2700672149658203,
      "learning_rate": 2.743171019658154e-06,
      "loss": 0.4741,
      "step": 4391
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.189936637878418,
      "learning_rate": 2.741000491065056e-06,
      "loss": 0.4903,
      "step": 4392
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3106061220169067,
      "learning_rate": 2.7388304972262776e-06,
      "loss": 0.4294,
      "step": 4393
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2138195037841797,
      "learning_rate": 2.7366610386555005e-06,
      "loss": 0.4163,
      "step": 4394
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6261540651321411,
      "learning_rate": 2.734492115866287e-06,
      "loss": 0.4839,
      "step": 4395
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.1247801780700684,
      "learning_rate": 2.732323729372064e-06,
      "loss": 0.4399,
      "step": 4396
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2323055267333984,
      "learning_rate": 2.7301558796861393e-06,
      "loss": 0.4757,
      "step": 4397
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1868032217025757,
      "learning_rate": 2.7279885673216876e-06,
      "loss": 0.4424,
      "step": 4398
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3532443046569824,
      "learning_rate": 2.7258217927917607e-06,
      "loss": 0.4949,
      "step": 4399
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4207618236541748,
      "learning_rate": 2.7236555566092814e-06,
      "loss": 0.5047,
      "step": 4400
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1710548400878906,
      "learning_rate": 2.721489859287041e-06,
      "loss": 0.3936,
      "step": 4401
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.9040530920028687,
      "learning_rate": 2.719324701337713e-06,
      "loss": 0.4988,
      "step": 4402
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5236027240753174,
      "learning_rate": 2.71716008327383e-06,
      "loss": 0.5531,
      "step": 4403
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1128789186477661,
      "learning_rate": 2.7149960056078123e-06,
      "loss": 0.7137,
      "step": 4404
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6987003087997437,
      "learning_rate": 2.7128324688519404e-06,
      "loss": 0.5157,
      "step": 4405
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3541868925094604,
      "learning_rate": 2.710669473518369e-06,
      "loss": 0.4389,
      "step": 4406
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.178709626197815,
      "learning_rate": 2.7085070201191297e-06,
      "loss": 0.4061,
      "step": 4407
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3348678350448608,
      "learning_rate": 2.7063451091661187e-06,
      "loss": 0.4502,
      "step": 4408
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.3950271606445312,
      "learning_rate": 2.70418374117111e-06,
      "loss": 0.4137,
      "step": 4409
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1611250638961792,
      "learning_rate": 2.702022916645744e-06,
      "loss": 0.4476,
      "step": 4410
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1583997011184692,
      "learning_rate": 2.699862636101538e-06,
      "loss": 0.6691,
      "step": 4411
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2036285400390625,
      "learning_rate": 2.6977029000498754e-06,
      "loss": 0.4313,
      "step": 4412
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4763379096984863,
      "learning_rate": 2.6955437090020108e-06,
      "loss": 0.4434,
      "step": 4413
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1000041961669922,
      "learning_rate": 2.6933850634690746e-06,
      "loss": 0.392,
      "step": 4414
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2108806371688843,
      "learning_rate": 2.6912269639620637e-06,
      "loss": 0.424,
      "step": 4415
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1603869199752808,
      "learning_rate": 2.68906941099185e-06,
      "loss": 0.5222,
      "step": 4416
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5364041328430176,
      "learning_rate": 2.686912405069171e-06,
      "loss": 0.4868,
      "step": 4417
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3530151844024658,
      "learning_rate": 2.6847559467046346e-06,
      "loss": 0.4502,
      "step": 4418
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1006507873535156,
      "learning_rate": 2.682600036408727e-06,
      "loss": 0.522,
      "step": 4419
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3960083723068237,
      "learning_rate": 2.6804446746917924e-06,
      "loss": 0.4562,
      "step": 4420
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4969874620437622,
      "learning_rate": 2.678289862064058e-06,
      "loss": 0.4722,
      "step": 4421
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3524543046951294,
      "learning_rate": 2.67613559903561e-06,
      "loss": 0.4279,
      "step": 4422
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1013835668563843,
      "learning_rate": 2.673981886116413e-06,
      "loss": 0.441,
      "step": 4423
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1790664196014404,
      "learning_rate": 2.6718287238162963e-06,
      "loss": 0.6618,
      "step": 4424
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1028426885604858,
      "learning_rate": 2.6696761126449557e-06,
      "loss": 0.4355,
      "step": 4425
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3072501420974731,
      "learning_rate": 2.6675240531119695e-06,
      "loss": 0.4244,
      "step": 4426
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2526187896728516,
      "learning_rate": 2.665372545726771e-06,
      "loss": 0.4705,
      "step": 4427
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4538086652755737,
      "learning_rate": 2.6632215909986725e-06,
      "loss": 0.5248,
      "step": 4428
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5894685983657837,
      "learning_rate": 2.6610711894368503e-06,
      "loss": 0.447,
      "step": 4429
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3062950372695923,
      "learning_rate": 2.658921341550349e-06,
      "loss": 0.4417,
      "step": 4430
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.291778326034546,
      "learning_rate": 2.6567720478480876e-06,
      "loss": 0.4881,
      "step": 4431
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4621304273605347,
      "learning_rate": 2.6546233088388475e-06,
      "loss": 0.5403,
      "step": 4432
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2305121421813965,
      "learning_rate": 2.6524751250312863e-06,
      "loss": 0.49,
      "step": 4433
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8670053482055664,
      "learning_rate": 2.6503274969339204e-06,
      "loss": 0.4783,
      "step": 4434
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3449900150299072,
      "learning_rate": 2.6481804250551456e-06,
      "loss": 0.4754,
      "step": 4435
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.434287667274475,
      "learning_rate": 2.6460339099032173e-06,
      "loss": 0.4587,
      "step": 4436
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.87607741355896,
      "learning_rate": 2.643887951986259e-06,
      "loss": 0.4841,
      "step": 4437
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.475051999092102,
      "learning_rate": 2.641742551812274e-06,
      "loss": 0.4078,
      "step": 4438
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5426632165908813,
      "learning_rate": 2.639597709889119e-06,
      "loss": 0.4438,
      "step": 4439
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4364030361175537,
      "learning_rate": 2.637453426724528e-06,
      "loss": 0.4435,
      "step": 4440
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1167491674423218,
      "learning_rate": 2.6353097028260975e-06,
      "loss": 0.4457,
      "step": 4441
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1358873844146729,
      "learning_rate": 2.633166538701293e-06,
      "loss": 0.4793,
      "step": 4442
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1594005823135376,
      "learning_rate": 2.6310239348574494e-06,
      "loss": 0.4875,
      "step": 4443
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4310966730117798,
      "learning_rate": 2.6288818918017667e-06,
      "loss": 0.5047,
      "step": 4444
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4355173110961914,
      "learning_rate": 2.6267404100413153e-06,
      "loss": 0.5046,
      "step": 4445
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1604700088500977,
      "learning_rate": 2.6245994900830265e-06,
      "loss": 0.4926,
      "step": 4446
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.238049030303955,
      "learning_rate": 2.6224591324337074e-06,
      "loss": 0.4713,
      "step": 4447
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2702655792236328,
      "learning_rate": 2.6203193376000223e-06,
      "loss": 0.5152,
      "step": 4448
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6528217792510986,
      "learning_rate": 2.6181801060885105e-06,
      "loss": 0.5506,
      "step": 4449
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5374666452407837,
      "learning_rate": 2.6160414384055753e-06,
      "loss": 0.4373,
      "step": 4450
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2376848459243774,
      "learning_rate": 2.6139033350574826e-06,
      "loss": 0.4423,
      "step": 4451
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.128745198249817,
      "learning_rate": 2.6117657965503712e-06,
      "loss": 0.4093,
      "step": 4452
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4887884855270386,
      "learning_rate": 2.609628823390242e-06,
      "loss": 0.4551,
      "step": 4453
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4148004055023193,
      "learning_rate": 2.6074924160829606e-06,
      "loss": 0.5197,
      "step": 4454
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.632140874862671,
      "learning_rate": 2.6053565751342647e-06,
      "loss": 0.5672,
      "step": 4455
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2239460945129395,
      "learning_rate": 2.6032213010497505e-06,
      "loss": 0.4584,
      "step": 4456
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.224578857421875,
      "learning_rate": 2.601086594334888e-06,
      "loss": 0.4411,
      "step": 4457
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3923577070236206,
      "learning_rate": 2.598952455495005e-06,
      "loss": 0.4719,
      "step": 4458
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5973010063171387,
      "learning_rate": 2.5968188850352993e-06,
      "loss": 0.4453,
      "step": 4459
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.133795142173767,
      "learning_rate": 2.594685883460837e-06,
      "loss": 0.7477,
      "step": 4460
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4030296802520752,
      "learning_rate": 2.5925534512765416e-06,
      "loss": 0.4303,
      "step": 4461
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2540373802185059,
      "learning_rate": 2.59042158898721e-06,
      "loss": 0.5307,
      "step": 4462
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4904770851135254,
      "learning_rate": 2.5882902970974953e-06,
      "loss": 0.446,
      "step": 4463
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2501013278961182,
      "learning_rate": 2.5861595761119263e-06,
      "loss": 0.43,
      "step": 4464
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2100207805633545,
      "learning_rate": 2.584029426534888e-06,
      "loss": 0.4952,
      "step": 4465
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2910946607589722,
      "learning_rate": 2.5818998488706314e-06,
      "loss": 0.4907,
      "step": 4466
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.107661247253418,
      "learning_rate": 2.5797708436232782e-06,
      "loss": 0.4071,
      "step": 4467
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.14803946018219,
      "learning_rate": 2.577642411296806e-06,
      "loss": 0.5599,
      "step": 4468
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1396920680999756,
      "learning_rate": 2.5755145523950655e-06,
      "loss": 0.4847,
      "step": 4469
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.411429524421692,
      "learning_rate": 2.5733872674217617e-06,
      "loss": 0.4931,
      "step": 4470
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3691482543945312,
      "learning_rate": 2.571260556880473e-06,
      "loss": 0.4513,
      "step": 4471
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1964231729507446,
      "learning_rate": 2.5691344212746385e-06,
      "loss": 0.4971,
      "step": 4472
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.307084083557129,
      "learning_rate": 2.567008861107558e-06,
      "loss": 0.5123,
      "step": 4473
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.255175232887268,
      "learning_rate": 2.564883876882401e-06,
      "loss": 0.3817,
      "step": 4474
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1229774951934814,
      "learning_rate": 2.5627594691021938e-06,
      "loss": 0.4376,
      "step": 4475
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1095689535140991,
      "learning_rate": 2.5606356382698338e-06,
      "loss": 0.6727,
      "step": 4476
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3290373086929321,
      "learning_rate": 2.5585123848880744e-06,
      "loss": 0.4248,
      "step": 4477
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7239274978637695,
      "learning_rate": 2.556389709459539e-06,
      "loss": 0.4961,
      "step": 4478
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.803117275238037,
      "learning_rate": 2.5542676124867103e-06,
      "loss": 0.5127,
      "step": 4479
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7704111337661743,
      "learning_rate": 2.5521460944719323e-06,
      "loss": 0.49,
      "step": 4480
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.453618288040161,
      "learning_rate": 2.5500251559174155e-06,
      "loss": 0.4523,
      "step": 4481
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.615358829498291,
      "learning_rate": 2.5479047973252345e-06,
      "loss": 0.4649,
      "step": 4482
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1357393264770508,
      "learning_rate": 2.545785019197322e-06,
      "loss": 0.5275,
      "step": 4483
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.386521577835083,
      "learning_rate": 2.5436658220354778e-06,
      "loss": 0.441,
      "step": 4484
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0390156507492065,
      "learning_rate": 2.541547206341358e-06,
      "loss": 0.4705,
      "step": 4485
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.3556416034698486,
      "learning_rate": 2.5394291726164898e-06,
      "loss": 0.4021,
      "step": 4486
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7137311697006226,
      "learning_rate": 2.537311721362254e-06,
      "loss": 0.5508,
      "step": 4487
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7152862548828125,
      "learning_rate": 2.5351948530799004e-06,
      "loss": 0.4719,
      "step": 4488
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0765324831008911,
      "learning_rate": 2.5330785682705344e-06,
      "loss": 0.5173,
      "step": 4489
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4621232748031616,
      "learning_rate": 2.5309628674351306e-06,
      "loss": 0.4257,
      "step": 4490
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.146375060081482,
      "learning_rate": 2.528847751074519e-06,
      "loss": 0.4956,
      "step": 4491
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1937793493270874,
      "learning_rate": 2.52673321968939e-06,
      "loss": 0.4232,
      "step": 4492
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0622183084487915,
      "learning_rate": 2.524619273780308e-06,
      "loss": 0.5263,
      "step": 4493
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.205887794494629,
      "learning_rate": 2.5225059138476845e-06,
      "loss": 0.44,
      "step": 4494
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.273069143295288,
      "learning_rate": 2.5203931403917957e-06,
      "loss": 0.4729,
      "step": 4495
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2346043586730957,
      "learning_rate": 2.518280953912787e-06,
      "loss": 0.7682,
      "step": 4496
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1944007873535156,
      "learning_rate": 2.516169354910653e-06,
      "loss": 0.4506,
      "step": 4497
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5301814079284668,
      "learning_rate": 2.5140583438852595e-06,
      "loss": 0.5613,
      "step": 4498
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6654566526412964,
      "learning_rate": 2.511947921336325e-06,
      "loss": 0.3954,
      "step": 4499
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6661919355392456,
      "learning_rate": 2.5098380877634366e-06,
      "loss": 0.5326,
      "step": 4500
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.134641408920288,
      "learning_rate": 2.507728843666034e-06,
      "loss": 0.5015,
      "step": 4501
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4056979417800903,
      "learning_rate": 2.5056201895434242e-06,
      "loss": 0.4942,
      "step": 4502
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.137360692024231,
      "learning_rate": 2.503512125894768e-06,
      "loss": 0.7237,
      "step": 4503
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2369352579116821,
      "learning_rate": 2.5014046532190916e-06,
      "loss": 0.4622,
      "step": 4504
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0638896226882935,
      "learning_rate": 2.499297772015282e-06,
      "loss": 0.723,
      "step": 4505
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.906496286392212,
      "learning_rate": 2.49719148278208e-06,
      "loss": 0.5183,
      "step": 4506
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.2986063957214355,
      "learning_rate": 2.4950857860180943e-06,
      "loss": 0.5431,
      "step": 4507
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2991364002227783,
      "learning_rate": 2.492980682221786e-06,
      "loss": 0.4536,
      "step": 4508
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2040750980377197,
      "learning_rate": 2.4908761718914783e-06,
      "loss": 0.5237,
      "step": 4509
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5409796237945557,
      "learning_rate": 2.4887722555253574e-06,
      "loss": 0.483,
      "step": 4510
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2921451330184937,
      "learning_rate": 2.4866689336214623e-06,
      "loss": 0.4149,
      "step": 4511
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5571249723434448,
      "learning_rate": 2.4845662066777e-06,
      "loss": 0.541,
      "step": 4512
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4760991334915161,
      "learning_rate": 2.482464075191826e-06,
      "loss": 0.4717,
      "step": 4513
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.137991428375244,
      "learning_rate": 2.4803625396614643e-06,
      "loss": 0.4296,
      "step": 4514
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.202048897743225,
      "learning_rate": 2.4782616005840954e-06,
      "loss": 0.3672,
      "step": 4515
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4888951778411865,
      "learning_rate": 2.4761612584570526e-06,
      "loss": 0.4729,
      "step": 4516
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.11554753780365,
      "learning_rate": 2.4740615137775375e-06,
      "loss": 0.4228,
      "step": 4517
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1099308729171753,
      "learning_rate": 2.471962367042601e-06,
      "loss": 0.4856,
      "step": 4518
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4920601844787598,
      "learning_rate": 2.46986381874916e-06,
      "loss": 0.5043,
      "step": 4519
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4480715990066528,
      "learning_rate": 2.4677658693939864e-06,
      "loss": 0.4808,
      "step": 4520
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.859086036682129,
      "learning_rate": 2.4656685194737068e-06,
      "loss": 0.4763,
      "step": 4521
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.209113597869873,
      "learning_rate": 2.4635717694848145e-06,
      "loss": 0.6895,
      "step": 4522
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1688531637191772,
      "learning_rate": 2.461475619923651e-06,
      "loss": 0.7264,
      "step": 4523
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.437721610069275,
      "learning_rate": 2.4593800712864254e-06,
      "loss": 0.4536,
      "step": 4524
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.084131956100464,
      "learning_rate": 2.4572851240691945e-06,
      "loss": 0.4524,
      "step": 4525
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8664072155952454,
      "learning_rate": 2.4551907787678807e-06,
      "loss": 0.483,
      "step": 4526
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3649287223815918,
      "learning_rate": 2.453097035878263e-06,
      "loss": 0.4989,
      "step": 4527
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3780033588409424,
      "learning_rate": 2.451003895895972e-06,
      "loss": 0.4797,
      "step": 4528
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3631932735443115,
      "learning_rate": 2.448911359316503e-06,
      "loss": 0.4441,
      "step": 4529
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2296104431152344,
      "learning_rate": 2.4468194266352014e-06,
      "loss": 0.4795,
      "step": 4530
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.161892890930176,
      "learning_rate": 2.4447280983472773e-06,
      "loss": 0.4958,
      "step": 4531
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9303181767463684,
      "learning_rate": 2.4426373749477904e-06,
      "loss": 0.4557,
      "step": 4532
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5448375940322876,
      "learning_rate": 2.4405472569316597e-06,
      "loss": 0.4905,
      "step": 4533
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1869233846664429,
      "learning_rate": 2.438457744793665e-06,
      "loss": 0.4326,
      "step": 4534
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4214284420013428,
      "learning_rate": 2.436368839028435e-06,
      "loss": 0.4297,
      "step": 4535
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2134209871292114,
      "learning_rate": 2.434280540130464e-06,
      "loss": 0.5109,
      "step": 4536
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2413108348846436,
      "learning_rate": 2.4321928485940937e-06,
      "loss": 0.4293,
      "step": 4537
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1235140562057495,
      "learning_rate": 2.4301057649135276e-06,
      "loss": 0.4043,
      "step": 4538
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3393229246139526,
      "learning_rate": 2.4280192895828254e-06,
      "loss": 0.5022,
      "step": 4539
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3407429456710815,
      "learning_rate": 2.4259334230958983e-06,
      "loss": 0.4631,
      "step": 4540
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3108502626419067,
      "learning_rate": 2.423848165946519e-06,
      "loss": 0.4803,
      "step": 4541
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.289042592048645,
      "learning_rate": 2.4217635186283103e-06,
      "loss": 0.4618,
      "step": 4542
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0919126272201538,
      "learning_rate": 2.419679481634757e-06,
      "loss": 0.4868,
      "step": 4543
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5798341035842896,
      "learning_rate": 2.4175960554591936e-06,
      "loss": 0.5781,
      "step": 4544
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2110437154769897,
      "learning_rate": 2.4155132405948117e-06,
      "loss": 0.4815,
      "step": 4545
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2450789213180542,
      "learning_rate": 2.413431037534662e-06,
      "loss": 0.4999,
      "step": 4546
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.177515983581543,
      "learning_rate": 2.411349446771643e-06,
      "loss": 0.4256,
      "step": 4547
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1868852376937866,
      "learning_rate": 2.4092684687985157e-06,
      "loss": 0.4336,
      "step": 4548
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8468955755233765,
      "learning_rate": 2.407188104107893e-06,
      "loss": 0.4527,
      "step": 4549
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.227755069732666,
      "learning_rate": 2.4051083531922402e-06,
      "loss": 0.4661,
      "step": 4550
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2860552072525024,
      "learning_rate": 2.403029216543883e-06,
      "loss": 0.4719,
      "step": 4551
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5091278553009033,
      "learning_rate": 2.4009506946549937e-06,
      "loss": 0.4344,
      "step": 4552
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0572328567504883,
      "learning_rate": 2.398872788017608e-06,
      "loss": 0.4746,
      "step": 4553
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5240764617919922,
      "learning_rate": 2.396795497123608e-06,
      "loss": 0.4749,
      "step": 4554
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1627209186553955,
      "learning_rate": 2.3947188224647373e-06,
      "loss": 0.4003,
      "step": 4555
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.133805274963379,
      "learning_rate": 2.392642764532588e-06,
      "loss": 0.4608,
      "step": 4556
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1944092512130737,
      "learning_rate": 2.3905673238186054e-06,
      "loss": 0.7387,
      "step": 4557
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.301126718521118,
      "learning_rate": 2.388492500814097e-06,
      "loss": 0.5103,
      "step": 4558
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3889707326889038,
      "learning_rate": 2.386418296010212e-06,
      "loss": 0.498,
      "step": 4559
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3458430767059326,
      "learning_rate": 2.384344709897967e-06,
      "loss": 0.5512,
      "step": 4560
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4058507680892944,
      "learning_rate": 2.382271742968221e-06,
      "loss": 0.4998,
      "step": 4561
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.214120864868164,
      "learning_rate": 2.38019939571169e-06,
      "loss": 0.4832,
      "step": 4562
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3801859617233276,
      "learning_rate": 2.3781276686189453e-06,
      "loss": 0.4579,
      "step": 4563
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2998695373535156,
      "learning_rate": 2.3760565621804076e-06,
      "loss": 0.4804,
      "step": 4564
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1660720109939575,
      "learning_rate": 2.373986076886356e-06,
      "loss": 0.5147,
      "step": 4565
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2507717609405518,
      "learning_rate": 2.371916213226916e-06,
      "loss": 0.5405,
      "step": 4566
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0014797449111938,
      "learning_rate": 2.3698469716920725e-06,
      "loss": 0.722,
      "step": 4567
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3451135158538818,
      "learning_rate": 2.3677783527716587e-06,
      "loss": 0.4732,
      "step": 4568
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3349273204803467,
      "learning_rate": 2.3657103569553595e-06,
      "loss": 0.5023,
      "step": 4569
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2684605121612549,
      "learning_rate": 2.3636429847327163e-06,
      "loss": 0.6901,
      "step": 4570
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3707116842269897,
      "learning_rate": 2.361576236593121e-06,
      "loss": 0.5,
      "step": 4571
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2201467752456665,
      "learning_rate": 2.3595101130258203e-06,
      "loss": 0.5073,
      "step": 4572
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3157845735549927,
      "learning_rate": 2.3574446145199083e-06,
      "loss": 0.4369,
      "step": 4573
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1198519468307495,
      "learning_rate": 2.3553797415643315e-06,
      "loss": 0.4469,
      "step": 4574
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3342493772506714,
      "learning_rate": 2.3533154946478935e-06,
      "loss": 0.5024,
      "step": 4575
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0717517137527466,
      "learning_rate": 2.3512518742592437e-06,
      "loss": 0.4961,
      "step": 4576
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2985775470733643,
      "learning_rate": 2.3491888808868894e-06,
      "loss": 0.3787,
      "step": 4577
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1752701997756958,
      "learning_rate": 2.347126515019182e-06,
      "loss": 0.4305,
      "step": 4578
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2825690507888794,
      "learning_rate": 2.3450647771443317e-06,
      "loss": 0.491,
      "step": 4579
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.92446768283844,
      "learning_rate": 2.343003667750396e-06,
      "loss": 0.3973,
      "step": 4580
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0561232566833496,
      "learning_rate": 2.3409431873252797e-06,
      "loss": 0.4059,
      "step": 4581
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1475825309753418,
      "learning_rate": 2.338883336356751e-06,
      "loss": 0.499,
      "step": 4582
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8688163757324219,
      "learning_rate": 2.3368241153324157e-06,
      "loss": 0.532,
      "step": 4583
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0626561641693115,
      "learning_rate": 2.33476552473974e-06,
      "loss": 0.4339,
      "step": 4584
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3301464319229126,
      "learning_rate": 2.332707565066036e-06,
      "loss": 0.4668,
      "step": 4585
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.475051760673523,
      "learning_rate": 2.330650236798464e-06,
      "loss": 0.4866,
      "step": 4586
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0711044073104858,
      "learning_rate": 2.3285935404240432e-06,
      "loss": 0.4428,
      "step": 4587
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5543972253799438,
      "learning_rate": 2.3265374764296343e-06,
      "loss": 0.44,
      "step": 4588
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.60267174243927,
      "learning_rate": 2.3244820453019566e-06,
      "loss": 0.4725,
      "step": 4589
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5679116249084473,
      "learning_rate": 2.322427247527571e-06,
      "loss": 0.4767,
      "step": 4590
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1628504991531372,
      "learning_rate": 2.3203730835928974e-06,
      "loss": 0.427,
      "step": 4591
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9613525867462158,
      "learning_rate": 2.3183195539841964e-06,
      "loss": 0.4433,
      "step": 4592
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.518041729927063,
      "learning_rate": 2.3162666591875852e-06,
      "loss": 0.4307,
      "step": 4593
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7784799337387085,
      "learning_rate": 2.314214399689031e-06,
      "loss": 0.4058,
      "step": 4594
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4076770544052124,
      "learning_rate": 2.3121627759743435e-06,
      "loss": 0.5362,
      "step": 4595
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.4106156826019287,
      "learning_rate": 2.310111788529192e-06,
      "loss": 0.5121,
      "step": 4596
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.666127324104309,
      "learning_rate": 2.308061437839087e-06,
      "loss": 0.4615,
      "step": 4597
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3772525787353516,
      "learning_rate": 2.306011724389389e-06,
      "loss": 0.4535,
      "step": 4598
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9637943506240845,
      "learning_rate": 2.3039626486653134e-06,
      "loss": 0.4139,
      "step": 4599
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0436123609542847,
      "learning_rate": 2.3019142111519182e-06,
      "loss": 0.4708,
      "step": 4600
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4752088785171509,
      "learning_rate": 2.2998664123341163e-06,
      "loss": 0.4736,
      "step": 4601
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1802945137023926,
      "learning_rate": 2.297819252696662e-06,
      "loss": 0.5028,
      "step": 4602
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3424638509750366,
      "learning_rate": 2.2957727327241673e-06,
      "loss": 0.4104,
      "step": 4603
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6345417499542236,
      "learning_rate": 2.2937268529010836e-06,
      "loss": 0.4065,
      "step": 4604
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1968315839767456,
      "learning_rate": 2.291681613711717e-06,
      "loss": 0.5025,
      "step": 4605
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6053407192230225,
      "learning_rate": 2.289637015640222e-06,
      "loss": 0.4706,
      "step": 4606
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.1461567878723145,
      "learning_rate": 2.2875930591705957e-06,
      "loss": 0.5023,
      "step": 4607
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2057119607925415,
      "learning_rate": 2.2855497447866916e-06,
      "loss": 0.487,
      "step": 4608
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.024658679962158,
      "learning_rate": 2.283507072972203e-06,
      "loss": 0.4669,
      "step": 4609
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.244205355644226,
      "learning_rate": 2.2814650442106744e-06,
      "loss": 0.4844,
      "step": 4610
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5009825229644775,
      "learning_rate": 2.279423658985502e-06,
      "loss": 0.4577,
      "step": 4611
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1973618268966675,
      "learning_rate": 2.277382917779922e-06,
      "loss": 0.4107,
      "step": 4612
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4747179746627808,
      "learning_rate": 2.2753428210770256e-06,
      "loss": 0.5323,
      "step": 4613
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1187283992767334,
      "learning_rate": 2.2733033693597452e-06,
      "loss": 0.5331,
      "step": 4614
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2476564645767212,
      "learning_rate": 2.2712645631108644e-06,
      "loss": 0.4689,
      "step": 4615
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2229365110397339,
      "learning_rate": 2.2692264028130158e-06,
      "loss": 0.5106,
      "step": 4616
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3919576406478882,
      "learning_rate": 2.2671888889486714e-06,
      "loss": 0.4192,
      "step": 4617
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8856934309005737,
      "learning_rate": 2.2651520220001594e-06,
      "loss": 0.4416,
      "step": 4618
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.521559476852417,
      "learning_rate": 2.263115802449647e-06,
      "loss": 0.4911,
      "step": 4619
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1905279159545898,
      "learning_rate": 2.261080230779155e-06,
      "loss": 0.4354,
      "step": 4620
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1283537149429321,
      "learning_rate": 2.259045307470544e-06,
      "loss": 0.4169,
      "step": 4621
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2025929689407349,
      "learning_rate": 2.2570110330055284e-06,
      "loss": 0.4647,
      "step": 4622
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0854259729385376,
      "learning_rate": 2.254977407865663e-06,
      "loss": 0.4719,
      "step": 4623
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1315065622329712,
      "learning_rate": 2.2529444325323495e-06,
      "loss": 0.426,
      "step": 4624
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3703199625015259,
      "learning_rate": 2.2509121074868405e-06,
      "loss": 0.5617,
      "step": 4625
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1396476030349731,
      "learning_rate": 2.248880433210229e-06,
      "loss": 0.386,
      "step": 4626
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3744325637817383,
      "learning_rate": 2.2468494101834577e-06,
      "loss": 0.5275,
      "step": 4627
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1973897218704224,
      "learning_rate": 2.244819038887315e-06,
      "loss": 0.4636,
      "step": 4628
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3939521312713623,
      "learning_rate": 2.242789319802432e-06,
      "loss": 0.5107,
      "step": 4629
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1121505498886108,
      "learning_rate": 2.2407602534092896e-06,
      "loss": 0.4515,
      "step": 4630
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.33594810962677,
      "learning_rate": 2.238731840188209e-06,
      "loss": 0.387,
      "step": 4631
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.026986122131348,
      "learning_rate": 2.236704080619363e-06,
      "loss": 0.4537,
      "step": 4632
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1723850965499878,
      "learning_rate": 2.234676975182763e-06,
      "loss": 0.4752,
      "step": 4633
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.2023773193359375,
      "learning_rate": 2.232650524358272e-06,
      "loss": 0.4259,
      "step": 4634
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3202698230743408,
      "learning_rate": 2.2306247286255938e-06,
      "loss": 0.7593,
      "step": 4635
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8115414381027222,
      "learning_rate": 2.2285995884642765e-06,
      "loss": 0.4857,
      "step": 4636
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.213127851486206,
      "learning_rate": 2.2265751043537166e-06,
      "loss": 0.5316,
      "step": 4637
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3477381467819214,
      "learning_rate": 2.224551276773155e-06,
      "loss": 0.4118,
      "step": 4638
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1910440921783447,
      "learning_rate": 2.222528106201672e-06,
      "loss": 0.4566,
      "step": 4639
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.165477752685547,
      "learning_rate": 2.2205055931182005e-06,
      "loss": 0.4549,
      "step": 4640
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2352579832077026,
      "learning_rate": 2.218483738001509e-06,
      "loss": 0.4024,
      "step": 4641
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3134875297546387,
      "learning_rate": 2.2164625413302186e-06,
      "loss": 0.4796,
      "step": 4642
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3918300867080688,
      "learning_rate": 2.214442003582786e-06,
      "loss": 0.5055,
      "step": 4643
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.265845775604248,
      "learning_rate": 2.2124221252375215e-06,
      "loss": 0.4879,
      "step": 4644
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2204068899154663,
      "learning_rate": 2.21040290677257e-06,
      "loss": 0.4129,
      "step": 4645
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2063313722610474,
      "learning_rate": 2.208384348665928e-06,
      "loss": 0.4049,
      "step": 4646
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1571531295776367,
      "learning_rate": 2.20636645139543e-06,
      "loss": 0.4291,
      "step": 4647
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3478389978408813,
      "learning_rate": 2.2043492154387518e-06,
      "loss": 0.5112,
      "step": 4648
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3645431995391846,
      "learning_rate": 2.2023326412734263e-06,
      "loss": 0.4732,
      "step": 4649
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3363311290740967,
      "learning_rate": 2.200316729376814e-06,
      "loss": 0.4235,
      "step": 4650
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.50580632686615,
      "learning_rate": 2.198301480226128e-06,
      "loss": 0.5399,
      "step": 4651
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3393144607543945,
      "learning_rate": 2.1962868942984213e-06,
      "loss": 0.4647,
      "step": 4652
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4207340478897095,
      "learning_rate": 2.1942729720705876e-06,
      "loss": 0.4367,
      "step": 4653
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2931928634643555,
      "learning_rate": 2.192259714019369e-06,
      "loss": 0.4769,
      "step": 4654
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0677672624588013,
      "learning_rate": 2.190247120621345e-06,
      "loss": 0.5139,
      "step": 4655
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4895216226577759,
      "learning_rate": 2.1882351923529432e-06,
      "loss": 0.4347,
      "step": 4656
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.235998511314392,
      "learning_rate": 2.1862239296904274e-06,
      "loss": 0.7794,
      "step": 4657
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2488529682159424,
      "learning_rate": 2.184213333109911e-06,
      "loss": 0.4238,
      "step": 4658
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6516236066818237,
      "learning_rate": 2.1822034030873424e-06,
      "loss": 0.4973,
      "step": 4659
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7743109464645386,
      "learning_rate": 2.1801941400985165e-06,
      "loss": 0.4666,
      "step": 4660
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1326850652694702,
      "learning_rate": 2.178185544619073e-06,
      "loss": 0.4532,
      "step": 4661
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4954168796539307,
      "learning_rate": 2.176177617124485e-06,
      "loss": 0.4507,
      "step": 4662
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6008880138397217,
      "learning_rate": 2.1741703580900774e-06,
      "loss": 0.4802,
      "step": 4663
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3114932775497437,
      "learning_rate": 2.1721637679910096e-06,
      "loss": 0.4511,
      "step": 4664
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.192803144454956,
      "learning_rate": 2.1701578473022833e-06,
      "loss": 0.4434,
      "step": 4665
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1841259002685547,
      "learning_rate": 2.1681525964987477e-06,
      "loss": 0.466,
      "step": 4666
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5072256326675415,
      "learning_rate": 2.166148016055085e-06,
      "loss": 0.4176,
      "step": 4667
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1166198253631592,
      "learning_rate": 2.1641441064458276e-06,
      "loss": 0.4306,
      "step": 4668
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1945924758911133,
      "learning_rate": 2.1621408681453395e-06,
      "loss": 0.4991,
      "step": 4669
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.216753363609314,
      "learning_rate": 2.1601383016278333e-06,
      "loss": 0.4609,
      "step": 4670
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1249282360076904,
      "learning_rate": 2.158136407367362e-06,
      "loss": 0.6874,
      "step": 4671
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2177726030349731,
      "learning_rate": 2.156135185837814e-06,
      "loss": 0.4493,
      "step": 4672
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5955923795700073,
      "learning_rate": 2.154134637512925e-06,
      "loss": 0.4163,
      "step": 4673
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0977931022644043,
      "learning_rate": 2.1521347628662655e-06,
      "loss": 0.6635,
      "step": 4674
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.457216501235962,
      "learning_rate": 2.150135562371252e-06,
      "loss": 0.4047,
      "step": 4675
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.47446870803833,
      "learning_rate": 2.1481370365011376e-06,
      "loss": 0.426,
      "step": 4676
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.22211754322052,
      "learning_rate": 2.146139185729015e-06,
      "loss": 0.4295,
      "step": 4677
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.148982048034668,
      "learning_rate": 2.1441420105278223e-06,
      "loss": 0.4585,
      "step": 4678
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3150373697280884,
      "learning_rate": 2.14214551137033e-06,
      "loss": 0.4845,
      "step": 4679
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0885705947875977,
      "learning_rate": 2.140149688729158e-06,
      "loss": 0.4909,
      "step": 4680
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.527726650238037,
      "learning_rate": 2.1381545430767565e-06,
      "loss": 0.4587,
      "step": 4681
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5874239206314087,
      "learning_rate": 2.1361600748854215e-06,
      "loss": 0.5753,
      "step": 4682
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2488224506378174,
      "learning_rate": 2.1341662846272892e-06,
      "loss": 0.4742,
      "step": 4683
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1077404022216797,
      "learning_rate": 2.1321731727743293e-06,
      "loss": 0.4514,
      "step": 4684
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5344502925872803,
      "learning_rate": 2.1301807397983587e-06,
      "loss": 0.4512,
      "step": 4685
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.234520435333252,
      "learning_rate": 2.128188986171026e-06,
      "loss": 0.447,
      "step": 4686
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1699126958847046,
      "learning_rate": 2.1261979123638253e-06,
      "loss": 0.4251,
      "step": 4687
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5987203121185303,
      "learning_rate": 2.124207518848086e-06,
      "loss": 0.4793,
      "step": 4688
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0750635862350464,
      "learning_rate": 2.122217806094976e-06,
      "loss": 0.4932,
      "step": 4689
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3409826755523682,
      "learning_rate": 2.120228774575507e-06,
      "loss": 0.4822,
      "step": 4690
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1911066770553589,
      "learning_rate": 2.1182404247605217e-06,
      "loss": 0.4114,
      "step": 4691
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1705132722854614,
      "learning_rate": 2.11625275712071e-06,
      "loss": 0.4047,
      "step": 4692
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8771944046020508,
      "learning_rate": 2.1142657721265926e-06,
      "loss": 0.3712,
      "step": 4693
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0793483257293701,
      "learning_rate": 2.1122794702485334e-06,
      "loss": 0.3957,
      "step": 4694
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.152579665184021,
      "learning_rate": 2.110293851956735e-06,
      "loss": 0.4369,
      "step": 4695
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.195533275604248,
      "learning_rate": 2.108308917721234e-06,
      "loss": 0.4648,
      "step": 4696
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2687530517578125,
      "learning_rate": 2.1063246680119095e-06,
      "loss": 0.4048,
      "step": 4697
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0791525840759277,
      "learning_rate": 2.104341103298474e-06,
      "loss": 0.4684,
      "step": 4698
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1515017747879028,
      "learning_rate": 2.1023582240504836e-06,
      "loss": 0.4188,
      "step": 4699
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.720259428024292,
      "learning_rate": 2.1003760307373273e-06,
      "loss": 0.4788,
      "step": 4700
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1168172359466553,
      "learning_rate": 2.0983945238282316e-06,
      "loss": 0.4868,
      "step": 4701
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1945555210113525,
      "learning_rate": 2.096413703792266e-06,
      "loss": 0.4181,
      "step": 4702
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2568511962890625,
      "learning_rate": 2.09443357109833e-06,
      "loss": 0.4835,
      "step": 4703
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.607810139656067,
      "learning_rate": 2.0924541262151655e-06,
      "loss": 0.497,
      "step": 4704
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1407748460769653,
      "learning_rate": 2.0904753696113524e-06,
      "loss": 0.7537,
      "step": 4705
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2365200519561768,
      "learning_rate": 2.0884973017553012e-06,
      "loss": 0.4599,
      "step": 4706
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4445056915283203,
      "learning_rate": 2.086519923115268e-06,
      "loss": 0.493,
      "step": 4707
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0903875827789307,
      "learning_rate": 2.0845432341593375e-06,
      "loss": 0.4175,
      "step": 4708
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4352909326553345,
      "learning_rate": 2.0825672353554383e-06,
      "loss": 0.4537,
      "step": 4709
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.243970513343811,
      "learning_rate": 2.0805919271713286e-06,
      "loss": 0.4647,
      "step": 4710
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0650160312652588,
      "learning_rate": 2.078617310074611e-06,
      "loss": 0.5127,
      "step": 4711
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3134303092956543,
      "learning_rate": 2.0766433845327185e-06,
      "loss": 0.5301,
      "step": 4712
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2491058111190796,
      "learning_rate": 2.07467015101292e-06,
      "loss": 0.4047,
      "step": 4713
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.182573914527893,
      "learning_rate": 2.072697609982326e-06,
      "loss": 0.4806,
      "step": 4714
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2584689855575562,
      "learning_rate": 2.0707257619078746e-06,
      "loss": 0.4752,
      "step": 4715
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5054371356964111,
      "learning_rate": 2.0687546072563534e-06,
      "loss": 0.4383,
      "step": 4716
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.844831109046936,
      "learning_rate": 2.0667841464943735e-06,
      "loss": 0.476,
      "step": 4717
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0947734117507935,
      "learning_rate": 2.064814380088383e-06,
      "loss": 0.4602,
      "step": 4718
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5457775592803955,
      "learning_rate": 2.0628453085046737e-06,
      "loss": 0.4959,
      "step": 4719
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4143798351287842,
      "learning_rate": 2.0608769322093626e-06,
      "loss": 0.441,
      "step": 4720
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2767572402954102,
      "learning_rate": 2.0589092516684116e-06,
      "loss": 0.4103,
      "step": 4721
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.252524495124817,
      "learning_rate": 2.05694226734761e-06,
      "loss": 0.3794,
      "step": 4722
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0353502035140991,
      "learning_rate": 2.054975979712588e-06,
      "loss": 0.5049,
      "step": 4723
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0833157300949097,
      "learning_rate": 2.053010389228809e-06,
      "loss": 0.5168,
      "step": 4724
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.343167781829834,
      "learning_rate": 2.0510454963615672e-06,
      "loss": 0.5273,
      "step": 4725
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.842531681060791,
      "learning_rate": 2.049081301575999e-06,
      "loss": 0.417,
      "step": 4726
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.251619577407837,
      "learning_rate": 2.047117805337071e-06,
      "loss": 0.4889,
      "step": 4727
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3569793701171875,
      "learning_rate": 2.045155008109587e-06,
      "loss": 0.4844,
      "step": 4728
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5053746700286865,
      "learning_rate": 2.0431929103581832e-06,
      "loss": 0.5405,
      "step": 4729
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0993868112564087,
      "learning_rate": 2.041231512547328e-06,
      "loss": 0.6962,
      "step": 4730
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2132134437561035,
      "learning_rate": 2.0392708151413302e-06,
      "loss": 0.5102,
      "step": 4731
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.16682767868042,
      "learning_rate": 2.037310818604328e-06,
      "loss": 0.4691,
      "step": 4732
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4110311269760132,
      "learning_rate": 2.0353515234002964e-06,
      "loss": 0.5045,
      "step": 4733
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1245827674865723,
      "learning_rate": 2.0333929299930406e-06,
      "loss": 0.4837,
      "step": 4734
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.378337025642395,
      "learning_rate": 2.031435038846206e-06,
      "loss": 0.4406,
      "step": 4735
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.133907437324524,
      "learning_rate": 2.0294778504232638e-06,
      "loss": 0.4779,
      "step": 4736
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2301770448684692,
      "learning_rate": 2.027521365187526e-06,
      "loss": 0.4672,
      "step": 4737
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.384507894515991,
      "learning_rate": 2.0255655836021353e-06,
      "loss": 0.3763,
      "step": 4738
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.297508955001831,
      "learning_rate": 2.023610506130066e-06,
      "loss": 0.4946,
      "step": 4739
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1953495740890503,
      "learning_rate": 2.02165613323413e-06,
      "loss": 0.4596,
      "step": 4740
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3613014221191406,
      "learning_rate": 2.019702465376968e-06,
      "loss": 0.4971,
      "step": 4741
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1334617137908936,
      "learning_rate": 2.0177495030210546e-06,
      "loss": 0.4743,
      "step": 4742
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.184807538986206,
      "learning_rate": 2.015797246628702e-06,
      "loss": 0.452,
      "step": 4743
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.273245096206665,
      "learning_rate": 2.0138456966620467e-06,
      "loss": 0.5434,
      "step": 4744
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8562067747116089,
      "learning_rate": 2.0118948535830683e-06,
      "loss": 0.5031,
      "step": 4745
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7336196899414062,
      "learning_rate": 2.00994471785357e-06,
      "loss": 0.4599,
      "step": 4746
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1943727731704712,
      "learning_rate": 2.007995289935194e-06,
      "loss": 0.5002,
      "step": 4747
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2577255964279175,
      "learning_rate": 2.0060465702894092e-06,
      "loss": 0.4931,
      "step": 4748
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2195842266082764,
      "learning_rate": 2.004098559377522e-06,
      "loss": 0.4503,
      "step": 4749
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2842868566513062,
      "learning_rate": 2.00215125766067e-06,
      "loss": 0.522,
      "step": 4750
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1105456352233887,
      "learning_rate": 2.000204665599819e-06,
      "loss": 0.4539,
      "step": 4751
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5480282306671143,
      "learning_rate": 1.998258783655774e-06,
      "loss": 0.4513,
      "step": 4752
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1976900100708008,
      "learning_rate": 1.9963136122891636e-06,
      "loss": 0.4968,
      "step": 4753
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.67739737033844,
      "learning_rate": 1.9943691519604523e-06,
      "loss": 0.4504,
      "step": 4754
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4577572345733643,
      "learning_rate": 1.9924254031299388e-06,
      "loss": 0.4351,
      "step": 4755
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1767483949661255,
      "learning_rate": 1.990482366257747e-06,
      "loss": 0.4453,
      "step": 4756
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1014759540557861,
      "learning_rate": 1.988540041803841e-06,
      "loss": 0.3749,
      "step": 4757
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4214552640914917,
      "learning_rate": 1.9865984302280062e-06,
      "loss": 0.4485,
      "step": 4758
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.647294759750366,
      "learning_rate": 1.984657531989866e-06,
      "loss": 0.3733,
      "step": 4759
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3807015419006348,
      "learning_rate": 1.982717347548876e-06,
      "loss": 0.4105,
      "step": 4760
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.224549651145935,
      "learning_rate": 1.980777877364316e-06,
      "loss": 0.4716,
      "step": 4761
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4789063930511475,
      "learning_rate": 1.978839121895304e-06,
      "loss": 0.5277,
      "step": 4762
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2077456712722778,
      "learning_rate": 1.9769010816007828e-06,
      "loss": 0.4384,
      "step": 4763
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1740539073944092,
      "learning_rate": 1.974963756939532e-06,
      "loss": 0.4135,
      "step": 4764
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3038443326950073,
      "learning_rate": 1.973027148370154e-06,
      "loss": 0.4332,
      "step": 4765
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4177825450897217,
      "learning_rate": 1.9710912563510915e-06,
      "loss": 0.4829,
      "step": 4766
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2871407270431519,
      "learning_rate": 1.9691560813406084e-06,
      "loss": 0.5121,
      "step": 4767
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5255522727966309,
      "learning_rate": 1.967221623796803e-06,
      "loss": 0.5201,
      "step": 4768
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.289953351020813,
      "learning_rate": 1.9652878841776056e-06,
      "loss": 0.4986,
      "step": 4769
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6031142473220825,
      "learning_rate": 1.963354862940771e-06,
      "loss": 0.461,
      "step": 4770
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1899309158325195,
      "learning_rate": 1.96142256054389e-06,
      "loss": 0.45,
      "step": 4771
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.050955057144165,
      "learning_rate": 1.9594909774443817e-06,
      "loss": 0.5173,
      "step": 4772
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6528071165084839,
      "learning_rate": 1.9575601140994904e-06,
      "loss": 0.4703,
      "step": 4773
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.0033047199249268,
      "learning_rate": 1.955629970966297e-06,
      "loss": 0.47,
      "step": 4774
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1188287734985352,
      "learning_rate": 1.9537005485017047e-06,
      "loss": 0.4612,
      "step": 4775
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.497971773147583,
      "learning_rate": 1.9517718471624534e-06,
      "loss": 0.4161,
      "step": 4776
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.421520948410034,
      "learning_rate": 1.9498438674051057e-06,
      "loss": 0.5335,
      "step": 4777
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1820234060287476,
      "learning_rate": 1.9479166096860586e-06,
      "loss": 0.4047,
      "step": 4778
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2304894924163818,
      "learning_rate": 1.945990074461535e-06,
      "loss": 0.4566,
      "step": 4779
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1516085863113403,
      "learning_rate": 1.9440642621875868e-06,
      "loss": 0.4702,
      "step": 4780
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0961304903030396,
      "learning_rate": 1.9421391733200973e-06,
      "loss": 0.66,
      "step": 4781
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5690083503723145,
      "learning_rate": 1.9402148083147744e-06,
      "loss": 0.4606,
      "step": 4782
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3153613805770874,
      "learning_rate": 1.938291167627159e-06,
      "loss": 0.5088,
      "step": 4783
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2415707111358643,
      "learning_rate": 1.9363682517126198e-06,
      "loss": 0.4569,
      "step": 4784
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1924878358840942,
      "learning_rate": 1.93444606102635e-06,
      "loss": 0.474,
      "step": 4785
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1375874280929565,
      "learning_rate": 1.9325245960233773e-06,
      "loss": 0.3967,
      "step": 4786
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2527965307235718,
      "learning_rate": 1.9306038571585507e-06,
      "loss": 0.5,
      "step": 4787
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0533866882324219,
      "learning_rate": 1.9286838448865543e-06,
      "loss": 0.3789,
      "step": 4788
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5640294551849365,
      "learning_rate": 1.9267645596618934e-06,
      "loss": 0.5537,
      "step": 4789
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1062285900115967,
      "learning_rate": 1.9248460019389077e-06,
      "loss": 0.4202,
      "step": 4790
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2370505332946777,
      "learning_rate": 1.92292817217176e-06,
      "loss": 0.5083,
      "step": 4791
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.119322657585144,
      "learning_rate": 1.921011070814441e-06,
      "loss": 0.4445,
      "step": 4792
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5819889307022095,
      "learning_rate": 1.9190946983207714e-06,
      "loss": 0.5035,
      "step": 4793
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3483556509017944,
      "learning_rate": 1.9171790551443988e-06,
      "loss": 0.3939,
      "step": 4794
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2159987688064575,
      "learning_rate": 1.9152641417387986e-06,
      "loss": 0.7086,
      "step": 4795
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2755420207977295,
      "learning_rate": 1.913349958557272e-06,
      "loss": 0.3909,
      "step": 4796
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7228730916976929,
      "learning_rate": 1.9114365060529443e-06,
      "loss": 0.4916,
      "step": 4797
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4483622312545776,
      "learning_rate": 1.909523784678776e-06,
      "loss": 0.5111,
      "step": 4798
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.702524185180664,
      "learning_rate": 1.907611794887546e-06,
      "loss": 0.4805,
      "step": 4799
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.582400321960449,
      "learning_rate": 1.9057005371318677e-06,
      "loss": 0.4798,
      "step": 4800
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0440386533737183,
      "learning_rate": 1.9037900118641739e-06,
      "loss": 0.6734,
      "step": 4801
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2342544794082642,
      "learning_rate": 1.9018802195367303e-06,
      "loss": 0.4913,
      "step": 4802
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2171971797943115,
      "learning_rate": 1.8999711606016253e-06,
      "loss": 0.5212,
      "step": 4803
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1069310903549194,
      "learning_rate": 1.8980628355107705e-06,
      "loss": 0.471,
      "step": 4804
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.239068627357483,
      "learning_rate": 1.8961552447159154e-06,
      "loss": 0.4849,
      "step": 4805
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.096360683441162,
      "learning_rate": 1.894248388668623e-06,
      "loss": 0.4455,
      "step": 4806
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.867384195327759,
      "learning_rate": 1.8923422678202908e-06,
      "loss": 0.4372,
      "step": 4807
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2531485557556152,
      "learning_rate": 1.8904368826221375e-06,
      "loss": 0.4758,
      "step": 4808
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2639877796173096,
      "learning_rate": 1.8885322335252076e-06,
      "loss": 0.4477,
      "step": 4809
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.319046974182129,
      "learning_rate": 1.8866283209803766e-06,
      "loss": 0.4736,
      "step": 4810
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1001957654953003,
      "learning_rate": 1.8847251454383374e-06,
      "loss": 0.5149,
      "step": 4811
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9104173183441162,
      "learning_rate": 1.882822707349618e-06,
      "loss": 0.4505,
      "step": 4812
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.296859860420227,
      "learning_rate": 1.8809210071645628e-06,
      "loss": 0.4793,
      "step": 4813
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6729646921157837,
      "learning_rate": 1.879020045333349e-06,
      "loss": 0.508,
      "step": 4814
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3038796186447144,
      "learning_rate": 1.8771198223059721e-06,
      "loss": 0.4269,
      "step": 4815
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1767699718475342,
      "learning_rate": 1.875220338532258e-06,
      "loss": 0.45,
      "step": 4816
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2843352556228638,
      "learning_rate": 1.8733215944618582e-06,
      "loss": 0.5183,
      "step": 4817
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4235692024230957,
      "learning_rate": 1.8714235905442418e-06,
      "loss": 0.4061,
      "step": 4818
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.281046986579895,
      "learning_rate": 1.8695263272287124e-06,
      "loss": 0.5187,
      "step": 4819
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4173038005828857,
      "learning_rate": 1.8676298049643915e-06,
      "loss": 0.4537,
      "step": 4820
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.555112361907959,
      "learning_rate": 1.8657340242002248e-06,
      "loss": 0.5343,
      "step": 4821
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.013047218322754,
      "learning_rate": 1.8638389853849887e-06,
      "loss": 0.4642,
      "step": 4822
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.465578317642212,
      "learning_rate": 1.8619446889672771e-06,
      "loss": 0.4665,
      "step": 4823
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4194607734680176,
      "learning_rate": 1.8600511353955143e-06,
      "loss": 0.4588,
      "step": 4824
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9881260395050049,
      "learning_rate": 1.8581583251179413e-06,
      "loss": 0.4064,
      "step": 4825
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9501375555992126,
      "learning_rate": 1.8562662585826303e-06,
      "loss": 0.4686,
      "step": 4826
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4404499530792236,
      "learning_rate": 1.8543749362374764e-06,
      "loss": 0.4751,
      "step": 4827
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4052114486694336,
      "learning_rate": 1.852484358530192e-06,
      "loss": 0.4666,
      "step": 4828
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5871295928955078,
      "learning_rate": 1.850594525908323e-06,
      "loss": 0.4457,
      "step": 4829
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.285718560218811,
      "learning_rate": 1.84870543881923e-06,
      "loss": 0.4538,
      "step": 4830
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3798314332962036,
      "learning_rate": 1.8468170977101036e-06,
      "loss": 0.5322,
      "step": 4831
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2504664659500122,
      "learning_rate": 1.844929503027954e-06,
      "loss": 0.401,
      "step": 4832
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1440496444702148,
      "learning_rate": 1.8430426552196145e-06,
      "loss": 0.7057,
      "step": 4833
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5608501434326172,
      "learning_rate": 1.841156554731746e-06,
      "loss": 0.4591,
      "step": 4834
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0393168926239014,
      "learning_rate": 1.8392712020108266e-06,
      "loss": 0.3929,
      "step": 4835
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5681087970733643,
      "learning_rate": 1.8373865975031636e-06,
      "loss": 0.4575,
      "step": 4836
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5781482458114624,
      "learning_rate": 1.8355027416548799e-06,
      "loss": 0.513,
      "step": 4837
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3460862636566162,
      "learning_rate": 1.8336196349119273e-06,
      "loss": 0.5161,
      "step": 4838
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2508913278579712,
      "learning_rate": 1.8317372777200798e-06,
      "loss": 0.4921,
      "step": 4839
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5391467809677124,
      "learning_rate": 1.8298556705249287e-06,
      "loss": 0.4902,
      "step": 4840
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3308125734329224,
      "learning_rate": 1.8279748137718951e-06,
      "loss": 0.3886,
      "step": 4841
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3455705642700195,
      "learning_rate": 1.8260947079062153e-06,
      "loss": 0.5489,
      "step": 4842
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.301498532295227,
      "learning_rate": 1.824215353372954e-06,
      "loss": 0.5098,
      "step": 4843
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2954835891723633,
      "learning_rate": 1.8223367506169943e-06,
      "loss": 0.5036,
      "step": 4844
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1734071969985962,
      "learning_rate": 1.8204589000830403e-06,
      "loss": 0.4441,
      "step": 4845
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2723983526229858,
      "learning_rate": 1.8185818022156238e-06,
      "loss": 0.4885,
      "step": 4846
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4066311120986938,
      "learning_rate": 1.816705457459091e-06,
      "loss": 0.4562,
      "step": 4847
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7362480163574219,
      "learning_rate": 1.8148298662576148e-06,
      "loss": 0.4755,
      "step": 4848
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.0683860778808594,
      "learning_rate": 1.8129550290551913e-06,
      "loss": 0.5082,
      "step": 4849
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2211861610412598,
      "learning_rate": 1.8110809462956302e-06,
      "loss": 0.4931,
      "step": 4850
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2398695945739746,
      "learning_rate": 1.8092076184225727e-06,
      "loss": 0.4736,
      "step": 4851
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1993186473846436,
      "learning_rate": 1.807335045879472e-06,
      "loss": 0.4419,
      "step": 4852
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3594483137130737,
      "learning_rate": 1.8054632291096103e-06,
      "loss": 0.463,
      "step": 4853
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2028571367263794,
      "learning_rate": 1.8035921685560836e-06,
      "loss": 0.518,
      "step": 4854
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9910398125648499,
      "learning_rate": 1.8017218646618169e-06,
      "loss": 0.4262,
      "step": 4855
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.237006425857544,
      "learning_rate": 1.7998523178695493e-06,
      "loss": 0.4971,
      "step": 4856
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0659371614456177,
      "learning_rate": 1.7979835286218418e-06,
      "loss": 0.7387,
      "step": 4857
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6675050258636475,
      "learning_rate": 1.796115497361081e-06,
      "loss": 0.413,
      "step": 4858
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2572214603424072,
      "learning_rate": 1.7942482245294669e-06,
      "loss": 0.4892,
      "step": 4859
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3872745037078857,
      "learning_rate": 1.792381710569025e-06,
      "loss": 0.4739,
      "step": 4860
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0969781875610352,
      "learning_rate": 1.7905159559216024e-06,
      "loss": 0.4603,
      "step": 4861
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3303251266479492,
      "learning_rate": 1.7886509610288595e-06,
      "loss": 0.4625,
      "step": 4862
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2474164962768555,
      "learning_rate": 1.7867867263322846e-06,
      "loss": 0.4699,
      "step": 4863
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.663021206855774,
      "learning_rate": 1.7849232522731797e-06,
      "loss": 0.4407,
      "step": 4864
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.057020664215088,
      "learning_rate": 1.783060539292673e-06,
      "loss": 0.4795,
      "step": 4865
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.096013069152832,
      "learning_rate": 1.7811985878317057e-06,
      "loss": 0.475,
      "step": 4866
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3483896255493164,
      "learning_rate": 1.7793373983310452e-06,
      "loss": 0.4615,
      "step": 4867
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4746474027633667,
      "learning_rate": 1.777476971231274e-06,
      "loss": 0.5172,
      "step": 4868
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0673171281814575,
      "learning_rate": 1.7756173069727943e-06,
      "loss": 0.478,
      "step": 4869
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.30828058719635,
      "learning_rate": 1.7737584059958323e-06,
      "loss": 0.4618,
      "step": 4870
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.552871584892273,
      "learning_rate": 1.7719002687404252e-06,
      "loss": 0.4743,
      "step": 4871
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.1659672260284424,
      "learning_rate": 1.770042895646441e-06,
      "loss": 0.4089,
      "step": 4872
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3431968688964844,
      "learning_rate": 1.7681862871535577e-06,
      "loss": 0.4126,
      "step": 4873
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5670615434646606,
      "learning_rate": 1.7663304437012725e-06,
      "loss": 0.488,
      "step": 4874
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.2871856689453125,
      "learning_rate": 1.7644753657289071e-06,
      "loss": 0.4889,
      "step": 4875
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2181655168533325,
      "learning_rate": 1.7626210536755961e-06,
      "loss": 0.4722,
      "step": 4876
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6159471273422241,
      "learning_rate": 1.7607675079802989e-06,
      "loss": 0.557,
      "step": 4877
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1596306562423706,
      "learning_rate": 1.7589147290817859e-06,
      "loss": 0.4029,
      "step": 4878
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1916733980178833,
      "learning_rate": 1.7570627174186543e-06,
      "loss": 0.5154,
      "step": 4879
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2511134147644043,
      "learning_rate": 1.7552114734293118e-06,
      "loss": 0.4476,
      "step": 4880
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5664193630218506,
      "learning_rate": 1.7533609975519916e-06,
      "loss": 0.5083,
      "step": 4881
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.663116216659546,
      "learning_rate": 1.7515112902247383e-06,
      "loss": 0.5167,
      "step": 4882
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3108272552490234,
      "learning_rate": 1.7496623518854195e-06,
      "loss": 0.4763,
      "step": 4883
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4352039098739624,
      "learning_rate": 1.747814182971721e-06,
      "loss": 0.5189,
      "step": 4884
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1709842681884766,
      "learning_rate": 1.745966783921142e-06,
      "loss": 0.4035,
      "step": 4885
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6835516691207886,
      "learning_rate": 1.7441201551710019e-06,
      "loss": 0.5676,
      "step": 4886
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2335089445114136,
      "learning_rate": 1.7422742971584395e-06,
      "loss": 0.4629,
      "step": 4887
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2913566827774048,
      "learning_rate": 1.7404292103204073e-06,
      "loss": 0.4771,
      "step": 4888
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6689430475234985,
      "learning_rate": 1.7385848950936811e-06,
      "loss": 0.4707,
      "step": 4889
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.668247103691101,
      "learning_rate": 1.7367413519148462e-06,
      "loss": 0.413,
      "step": 4890
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5140265226364136,
      "learning_rate": 1.7348985812203128e-06,
      "loss": 0.4106,
      "step": 4891
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6028691530227661,
      "learning_rate": 1.7330565834463026e-06,
      "loss": 0.5238,
      "step": 4892
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7036398649215698,
      "learning_rate": 1.7312153590288567e-06,
      "loss": 0.4475,
      "step": 4893
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1319814920425415,
      "learning_rate": 1.7293749084038352e-06,
      "loss": 0.5076,
      "step": 4894
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8988386392593384,
      "learning_rate": 1.72753523200691e-06,
      "loss": 0.4829,
      "step": 4895
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.469857931137085,
      "learning_rate": 1.7256963302735752e-06,
      "loss": 0.445,
      "step": 4896
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3953534364700317,
      "learning_rate": 1.7238582036391372e-06,
      "loss": 0.5424,
      "step": 4897
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3367316722869873,
      "learning_rate": 1.7220208525387193e-06,
      "loss": 0.4844,
      "step": 4898
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3528120517730713,
      "learning_rate": 1.7201842774072652e-06,
      "loss": 0.4857,
      "step": 4899
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3598707914352417,
      "learning_rate": 1.7183484786795297e-06,
      "loss": 0.3964,
      "step": 4900
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3064137697219849,
      "learning_rate": 1.7165134567900887e-06,
      "loss": 0.5507,
      "step": 4901
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1534852981567383,
      "learning_rate": 1.7146792121733297e-06,
      "loss": 0.3801,
      "step": 4902
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0957307815551758,
      "learning_rate": 1.71284574526346e-06,
      "loss": 0.4088,
      "step": 4903
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2201204299926758,
      "learning_rate": 1.7110130564944988e-06,
      "loss": 0.4751,
      "step": 4904
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2443654537200928,
      "learning_rate": 1.709181146300285e-06,
      "loss": 0.5019,
      "step": 4905
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.758741855621338,
      "learning_rate": 1.707350015114473e-06,
      "loss": 0.4596,
      "step": 4906
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6246943473815918,
      "learning_rate": 1.7055196633705279e-06,
      "loss": 0.4544,
      "step": 4907
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7149906158447266,
      "learning_rate": 1.7036900915017379e-06,
      "loss": 0.4917,
      "step": 4908
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6505742073059082,
      "learning_rate": 1.701861299941198e-06,
      "loss": 0.4651,
      "step": 4909
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9735153913497925,
      "learning_rate": 1.7000332891218275e-06,
      "loss": 0.4153,
      "step": 4910
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2279599905014038,
      "learning_rate": 1.698206059476354e-06,
      "loss": 0.4263,
      "step": 4911
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1518099308013916,
      "learning_rate": 1.69637961143732e-06,
      "loss": 0.4462,
      "step": 4912
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3485122919082642,
      "learning_rate": 1.6945539454370901e-06,
      "loss": 0.4796,
      "step": 4913
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2448389530181885,
      "learning_rate": 1.6927290619078357e-06,
      "loss": 0.3812,
      "step": 4914
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.481865406036377,
      "learning_rate": 1.6909049612815475e-06,
      "loss": 0.4641,
      "step": 4915
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.297061800956726,
      "learning_rate": 1.6890816439900314e-06,
      "loss": 0.4516,
      "step": 4916
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4405429363250732,
      "learning_rate": 1.6872591104649038e-06,
      "loss": 0.4195,
      "step": 4917
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0273288488388062,
      "learning_rate": 1.6854373611376006e-06,
      "loss": 0.4092,
      "step": 4918
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6255545616149902,
      "learning_rate": 1.6836163964393664e-06,
      "loss": 0.4577,
      "step": 4919
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6696397066116333,
      "learning_rate": 1.6817962168012665e-06,
      "loss": 0.4605,
      "step": 4920
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4923943281173706,
      "learning_rate": 1.6799768226541735e-06,
      "loss": 0.46,
      "step": 4921
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2344048023223877,
      "learning_rate": 1.6781582144287811e-06,
      "loss": 0.707,
      "step": 4922
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1377341747283936,
      "learning_rate": 1.6763403925555915e-06,
      "loss": 0.5255,
      "step": 4923
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1297144889831543,
      "learning_rate": 1.674523357464921e-06,
      "loss": 0.4963,
      "step": 4924
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5000361204147339,
      "learning_rate": 1.6727071095869053e-06,
      "loss": 0.444,
      "step": 4925
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4085360765457153,
      "learning_rate": 1.670891649351486e-06,
      "loss": 0.455,
      "step": 4926
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.3829610347747803,
      "learning_rate": 1.6690769771884236e-06,
      "loss": 0.4852,
      "step": 4927
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2778202295303345,
      "learning_rate": 1.6672630935272922e-06,
      "loss": 0.516,
      "step": 4928
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3904637098312378,
      "learning_rate": 1.6654499987974742e-06,
      "loss": 0.4337,
      "step": 4929
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6363452672958374,
      "learning_rate": 1.6636376934281716e-06,
      "loss": 0.4526,
      "step": 4930
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3966823816299438,
      "learning_rate": 1.661826177848394e-06,
      "loss": 0.3857,
      "step": 4931
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4689562320709229,
      "learning_rate": 1.6600154524869693e-06,
      "loss": 0.4936,
      "step": 4932
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4467332363128662,
      "learning_rate": 1.658205517772532e-06,
      "loss": 0.5505,
      "step": 4933
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2044557332992554,
      "learning_rate": 1.6563963741335366e-06,
      "loss": 0.4249,
      "step": 4934
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2183201313018799,
      "learning_rate": 1.6545880219982451e-06,
      "loss": 0.4677,
      "step": 4935
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3691587448120117,
      "learning_rate": 1.6527804617947323e-06,
      "loss": 0.4463,
      "step": 4936
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.468029260635376,
      "learning_rate": 1.6509736939508886e-06,
      "loss": 0.4365,
      "step": 4937
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.183891773223877,
      "learning_rate": 1.6491677188944156e-06,
      "loss": 0.4722,
      "step": 4938
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.880070447921753,
      "learning_rate": 1.6473625370528274e-06,
      "loss": 0.4985,
      "step": 4939
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.124127984046936,
      "learning_rate": 1.645558148853449e-06,
      "loss": 0.3951,
      "step": 4940
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9455593824386597,
      "learning_rate": 1.6437545547234163e-06,
      "loss": 0.6907,
      "step": 4941
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.5208892822265625,
      "learning_rate": 1.641951755089683e-06,
      "loss": 0.4487,
      "step": 4942
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3498625755310059,
      "learning_rate": 1.6401497503790077e-06,
      "loss": 0.5154,
      "step": 4943
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5687072277069092,
      "learning_rate": 1.6383485410179667e-06,
      "loss": 0.4738,
      "step": 4944
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1560020446777344,
      "learning_rate": 1.6365481274329432e-06,
      "loss": 0.4539,
      "step": 4945
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2779752016067505,
      "learning_rate": 1.6347485100501365e-06,
      "loss": 0.5042,
      "step": 4946
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.09848952293396,
      "learning_rate": 1.6329496892955544e-06,
      "loss": 0.4594,
      "step": 4947
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0455659627914429,
      "learning_rate": 1.6311516655950144e-06,
      "loss": 0.456,
      "step": 4948
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2788490056991577,
      "learning_rate": 1.6293544393741506e-06,
      "loss": 0.4726,
      "step": 4949
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3218969106674194,
      "learning_rate": 1.6275580110584043e-06,
      "loss": 0.4303,
      "step": 4950
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2353190183639526,
      "learning_rate": 1.6257623810730316e-06,
      "loss": 0.4456,
      "step": 4951
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2072333097457886,
      "learning_rate": 1.623967549843095e-06,
      "loss": 0.4704,
      "step": 4952
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1143743991851807,
      "learning_rate": 1.6221735177934688e-06,
      "loss": 0.4016,
      "step": 4953
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5652698278427124,
      "learning_rate": 1.6203802853488426e-06,
      "loss": 0.5113,
      "step": 4954
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.407711386680603,
      "learning_rate": 1.6185878529337107e-06,
      "loss": 0.4113,
      "step": 4955
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4128614664077759,
      "learning_rate": 1.616796220972383e-06,
      "loss": 0.5318,
      "step": 4956
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3905446529388428,
      "learning_rate": 1.6150053898889756e-06,
      "loss": 0.452,
      "step": 4957
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.25652015209198,
      "learning_rate": 1.6132153601074202e-06,
      "loss": 0.4528,
      "step": 4958
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4858953952789307,
      "learning_rate": 1.6114261320514535e-06,
      "loss": 0.469,
      "step": 4959
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.262574553489685,
      "learning_rate": 1.6096377061446222e-06,
      "loss": 0.4582,
      "step": 4960
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.191606044769287,
      "learning_rate": 1.6078500828102917e-06,
      "loss": 0.4827,
      "step": 4961
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1833781003952026,
      "learning_rate": 1.6060632624716266e-06,
      "loss": 0.4474,
      "step": 4962
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.448390245437622,
      "learning_rate": 1.6042772455516092e-06,
      "loss": 0.4775,
      "step": 4963
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5243889093399048,
      "learning_rate": 1.6024920324730275e-06,
      "loss": 0.4204,
      "step": 4964
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.23454749584198,
      "learning_rate": 1.600707623658478e-06,
      "loss": 0.4465,
      "step": 4965
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2397160530090332,
      "learning_rate": 1.5989240195303723e-06,
      "loss": 0.4279,
      "step": 4966
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4545334577560425,
      "learning_rate": 1.5971412205109256e-06,
      "loss": 0.437,
      "step": 4967
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2056061029434204,
      "learning_rate": 1.595359227022168e-06,
      "loss": 0.4437,
      "step": 4968
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3300654888153076,
      "learning_rate": 1.593578039485933e-06,
      "loss": 0.4441,
      "step": 4969
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1840438842773438,
      "learning_rate": 1.5917976583238693e-06,
      "loss": 0.4315,
      "step": 4970
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.100329875946045,
      "learning_rate": 1.5900180839574298e-06,
      "loss": 0.4778,
      "step": 4971
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2857983112335205,
      "learning_rate": 1.5882393168078786e-06,
      "loss": 0.5225,
      "step": 4972
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.445920705795288,
      "learning_rate": 1.5864613572962912e-06,
      "loss": 0.5068,
      "step": 4973
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2737942934036255,
      "learning_rate": 1.5846842058435457e-06,
      "loss": 0.4388,
      "step": 4974
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1207680702209473,
      "learning_rate": 1.582907862870336e-06,
      "loss": 0.4158,
      "step": 4975
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5618218183517456,
      "learning_rate": 1.5811323287971585e-06,
      "loss": 0.4187,
      "step": 4976
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9724149703979492,
      "learning_rate": 1.5793576040443204e-06,
      "loss": 0.4496,
      "step": 4977
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.242187738418579,
      "learning_rate": 1.5775836890319407e-06,
      "loss": 0.4342,
      "step": 4978
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2599587440490723,
      "learning_rate": 1.5758105841799393e-06,
      "loss": 0.4778,
      "step": 4979
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9247632622718811,
      "learning_rate": 1.5740382899080532e-06,
      "loss": 0.3941,
      "step": 4980
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9884957075119019,
      "learning_rate": 1.5722668066358194e-06,
      "loss": 0.4089,
      "step": 4981
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2295900583267212,
      "learning_rate": 1.5704961347825875e-06,
      "loss": 0.5089,
      "step": 4982
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.915346086025238,
      "learning_rate": 1.568726274767517e-06,
      "loss": 0.3888,
      "step": 4983
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2080498933792114,
      "learning_rate": 1.5669572270095674e-06,
      "loss": 0.7014,
      "step": 4984
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0744843482971191,
      "learning_rate": 1.5651889919275153e-06,
      "loss": 0.7218,
      "step": 4985
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3843157291412354,
      "learning_rate": 1.5634215699399363e-06,
      "loss": 0.4586,
      "step": 4986
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3609949350357056,
      "learning_rate": 1.5616549614652216e-06,
      "loss": 0.4536,
      "step": 4987
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6304798126220703,
      "learning_rate": 1.5598891669215638e-06,
      "loss": 0.4722,
      "step": 4988
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.238986611366272,
      "learning_rate": 1.558124186726963e-06,
      "loss": 0.4001,
      "step": 4989
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.425803780555725,
      "learning_rate": 1.5563600212992326e-06,
      "loss": 0.4273,
      "step": 4990
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.223021149635315,
      "learning_rate": 1.5545966710559846e-06,
      "loss": 0.391,
      "step": 4991
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.201344609260559,
      "learning_rate": 1.5528341364146465e-06,
      "loss": 0.4851,
      "step": 4992
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5655198097229004,
      "learning_rate": 1.5510724177924447e-06,
      "loss": 0.5001,
      "step": 4993
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1748193502426147,
      "learning_rate": 1.549311515606418e-06,
      "loss": 0.4652,
      "step": 4994
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.24018394947052,
      "learning_rate": 1.5475514302734119e-06,
      "loss": 0.4975,
      "step": 4995
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7171014547348022,
      "learning_rate": 1.5457921622100742e-06,
      "loss": 0.4974,
      "step": 4996
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3047363758087158,
      "learning_rate": 1.5440337118328646e-06,
      "loss": 0.4544,
      "step": 4997
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2657876014709473,
      "learning_rate": 1.5422760795580433e-06,
      "loss": 0.4417,
      "step": 4998
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2631677389144897,
      "learning_rate": 1.5405192658016832e-06,
      "loss": 0.4233,
      "step": 4999
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3804786205291748,
      "learning_rate": 1.5387632709796596e-06,
      "loss": 0.4327,
      "step": 5000
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5498932600021362,
      "learning_rate": 1.5370080955076521e-06,
      "loss": 0.471,
      "step": 5001
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3778564929962158,
      "learning_rate": 1.5352537398011525e-06,
      "loss": 0.4898,
      "step": 5002
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1849257946014404,
      "learning_rate": 1.5335002042754516e-06,
      "loss": 0.4356,
      "step": 5003
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4574342966079712,
      "learning_rate": 1.5317474893456507e-06,
      "loss": 0.4822,
      "step": 5004
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5471149682998657,
      "learning_rate": 1.5299955954266583e-06,
      "loss": 0.5014,
      "step": 5005
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2648078203201294,
      "learning_rate": 1.5282445229331817e-06,
      "loss": 0.5037,
      "step": 5006
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4305675029754639,
      "learning_rate": 1.5264942722797404e-06,
      "loss": 0.4819,
      "step": 5007
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.116726040840149,
      "learning_rate": 1.524744843880655e-06,
      "loss": 0.4282,
      "step": 5008
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1688308715820312,
      "learning_rate": 1.5229962381500563e-06,
      "loss": 0.443,
      "step": 5009
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2589375972747803,
      "learning_rate": 1.521248455501873e-06,
      "loss": 0.3593,
      "step": 5010
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.170531988143921,
      "learning_rate": 1.5195014963498472e-06,
      "loss": 0.6582,
      "step": 5011
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0920376777648926,
      "learning_rate": 1.5177553611075207e-06,
      "loss": 0.4485,
      "step": 5012
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4040484428405762,
      "learning_rate": 1.5160100501882397e-06,
      "loss": 0.4414,
      "step": 5013
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4080389738082886,
      "learning_rate": 1.5142655640051602e-06,
      "loss": 0.4806,
      "step": 5014
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1336565017700195,
      "learning_rate": 1.5125219029712363e-06,
      "loss": 0.4401,
      "step": 5015
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.381548523902893,
      "learning_rate": 1.5107790674992357e-06,
      "loss": 0.4932,
      "step": 5016
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3705625534057617,
      "learning_rate": 1.5090370580017233e-06,
      "loss": 0.4489,
      "step": 5017
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0215405225753784,
      "learning_rate": 1.5072958748910677e-06,
      "loss": 0.4178,
      "step": 5018
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9025182723999023,
      "learning_rate": 1.5055555185794496e-06,
      "loss": 0.433,
      "step": 5019
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1198822259902954,
      "learning_rate": 1.5038159894788446e-06,
      "loss": 0.5027,
      "step": 5020
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1625243425369263,
      "learning_rate": 1.5020772880010414e-06,
      "loss": 0.3915,
      "step": 5021
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.299001932144165,
      "learning_rate": 1.5003394145576245e-06,
      "loss": 0.4798,
      "step": 5022
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.206428050994873,
      "learning_rate": 1.49860236955999e-06,
      "loss": 0.4288,
      "step": 5023
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4315061569213867,
      "learning_rate": 1.4968661534193308e-06,
      "loss": 0.4675,
      "step": 5024
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7228695154190063,
      "learning_rate": 1.4951307665466502e-06,
      "loss": 0.5353,
      "step": 5025
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.0308268070220947,
      "learning_rate": 1.4933962093527503e-06,
      "loss": 0.4578,
      "step": 5026
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5179543495178223,
      "learning_rate": 1.4916624822482356e-06,
      "loss": 0.4321,
      "step": 5027
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3325508832931519,
      "learning_rate": 1.489929585643523e-06,
      "loss": 0.4803,
      "step": 5028
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.694713830947876,
      "learning_rate": 1.4881975199488247e-06,
      "loss": 0.5203,
      "step": 5029
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2167028188705444,
      "learning_rate": 1.4864662855741552e-06,
      "loss": 0.4782,
      "step": 5030
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.168144702911377,
      "learning_rate": 1.4847358829293397e-06,
      "loss": 0.4803,
      "step": 5031
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3058134317398071,
      "learning_rate": 1.4830063124239979e-06,
      "loss": 0.3959,
      "step": 5032
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.268979549407959,
      "learning_rate": 1.4812775744675612e-06,
      "loss": 0.4494,
      "step": 5033
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.34237539768219,
      "learning_rate": 1.4795496694692557e-06,
      "loss": 0.5093,
      "step": 5034
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4151912927627563,
      "learning_rate": 1.4778225978381173e-06,
      "loss": 0.4784,
      "step": 5035
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.7348108291625977,
      "learning_rate": 1.4760963599829775e-06,
      "loss": 0.5133,
      "step": 5036
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.320263385772705,
      "learning_rate": 1.4743709563124786e-06,
      "loss": 0.4302,
      "step": 5037
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2147587537765503,
      "learning_rate": 1.4726463872350577e-06,
      "loss": 0.4883,
      "step": 5038
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3588464260101318,
      "learning_rate": 1.4709226531589594e-06,
      "loss": 0.4845,
      "step": 5039
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.130467414855957,
      "learning_rate": 1.4691997544922303e-06,
      "loss": 0.418,
      "step": 5040
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.925002574920654,
      "learning_rate": 1.4674776916427163e-06,
      "loss": 0.5428,
      "step": 5041
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2594146728515625,
      "learning_rate": 1.4657564650180667e-06,
      "loss": 0.4486,
      "step": 5042
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4021745920181274,
      "learning_rate": 1.464036075025735e-06,
      "loss": 0.5231,
      "step": 5043
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4712799787521362,
      "learning_rate": 1.462316522072973e-06,
      "loss": 0.4965,
      "step": 5044
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.09231698513031,
      "learning_rate": 1.460597806566838e-06,
      "loss": 0.3712,
      "step": 5045
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3094611167907715,
      "learning_rate": 1.4588799289141859e-06,
      "loss": 0.4543,
      "step": 5046
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.127814769744873,
      "learning_rate": 1.4571628895216778e-06,
      "loss": 0.4331,
      "step": 5047
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4707508087158203,
      "learning_rate": 1.4554466887957708e-06,
      "loss": 0.466,
      "step": 5048
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.252321720123291,
      "learning_rate": 1.4537313271427295e-06,
      "loss": 0.3983,
      "step": 5049
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.185544729232788,
      "learning_rate": 1.4520168049686184e-06,
      "loss": 0.4528,
      "step": 5050
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0451945066452026,
      "learning_rate": 1.4503031226792986e-06,
      "loss": 0.412,
      "step": 5051
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2577425241470337,
      "learning_rate": 1.4485902806804403e-06,
      "loss": 0.6817,
      "step": 5052
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.171705961227417,
      "learning_rate": 1.446878279377506e-06,
      "loss": 0.4808,
      "step": 5053
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2648870944976807,
      "learning_rate": 1.4451671191757676e-06,
      "loss": 0.5134,
      "step": 5054
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3212693929672241,
      "learning_rate": 1.4434568004802918e-06,
      "loss": 0.4973,
      "step": 5055
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1982566118240356,
      "learning_rate": 1.4417473236959478e-06,
      "loss": 0.368,
      "step": 5056
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3703739643096924,
      "learning_rate": 1.4400386892274076e-06,
      "loss": 0.4622,
      "step": 5057
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4576168060302734,
      "learning_rate": 1.4383308974791399e-06,
      "loss": 0.4354,
      "step": 5058
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2745603322982788,
      "learning_rate": 1.436623948855419e-06,
      "loss": 0.6978,
      "step": 5059
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2355022430419922,
      "learning_rate": 1.434917843760314e-06,
      "loss": 0.4994,
      "step": 5060
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5324078798294067,
      "learning_rate": 1.4332125825976984e-06,
      "loss": 0.4026,
      "step": 5061
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9062401056289673,
      "learning_rate": 1.431508165771246e-06,
      "loss": 0.4938,
      "step": 5062
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2893372774124146,
      "learning_rate": 1.4298045936844262e-06,
      "loss": 0.497,
      "step": 5063
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6131446361541748,
      "learning_rate": 1.428101866740515e-06,
      "loss": 0.4607,
      "step": 5064
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3089547157287598,
      "learning_rate": 1.4263999853425815e-06,
      "loss": 0.4279,
      "step": 5065
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.475704550743103,
      "learning_rate": 1.424698949893501e-06,
      "loss": 0.488,
      "step": 5066
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0910205841064453,
      "learning_rate": 1.422998760795945e-06,
      "loss": 0.4632,
      "step": 5067
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1048505306243896,
      "learning_rate": 1.4212994184523826e-06,
      "loss": 0.4448,
      "step": 5068
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0748820304870605,
      "learning_rate": 1.4196009232650887e-06,
      "loss": 0.5012,
      "step": 5069
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.243145227432251,
      "learning_rate": 1.4179032756361306e-06,
      "loss": 0.4933,
      "step": 5070
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3590915203094482,
      "learning_rate": 1.4162064759673804e-06,
      "loss": 0.4879,
      "step": 5071
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.935891032218933,
      "learning_rate": 1.414510524660509e-06,
      "loss": 0.5174,
      "step": 5072
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0883257389068604,
      "learning_rate": 1.4128154221169815e-06,
      "loss": 0.4675,
      "step": 5073
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1969075202941895,
      "learning_rate": 1.4111211687380688e-06,
      "loss": 0.4634,
      "step": 5074
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1616896390914917,
      "learning_rate": 1.4094277649248345e-06,
      "loss": 0.7144,
      "step": 5075
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1978760957717896,
      "learning_rate": 1.4077352110781473e-06,
      "loss": 0.456,
      "step": 5076
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.492412805557251,
      "learning_rate": 1.4060435075986684e-06,
      "loss": 0.5257,
      "step": 5077
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4590188264846802,
      "learning_rate": 1.404352654886864e-06,
      "loss": 0.4966,
      "step": 5078
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1766526699066162,
      "learning_rate": 1.4026626533429938e-06,
      "loss": 0.4695,
      "step": 5079
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.381431221961975,
      "learning_rate": 1.4009735033671163e-06,
      "loss": 0.4579,
      "step": 5080
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0951812267303467,
      "learning_rate": 1.3992852053590938e-06,
      "loss": 0.6709,
      "step": 5081
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1432814598083496,
      "learning_rate": 1.397597759718578e-06,
      "loss": 0.3961,
      "step": 5082
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.72367262840271,
      "learning_rate": 1.3959111668450305e-06,
      "loss": 0.4466,
      "step": 5083
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2568074464797974,
      "learning_rate": 1.3942254271377004e-06,
      "loss": 0.5291,
      "step": 5084
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7642751932144165,
      "learning_rate": 1.3925405409956388e-06,
      "loss": 0.4976,
      "step": 5085
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2119953632354736,
      "learning_rate": 1.3908565088176973e-06,
      "loss": 0.4862,
      "step": 5086
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2898000478744507,
      "learning_rate": 1.3891733310025202e-06,
      "loss": 0.4897,
      "step": 5087
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9530221819877625,
      "learning_rate": 1.3874910079485549e-06,
      "loss": 0.4615,
      "step": 5088
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4879672527313232,
      "learning_rate": 1.3858095400540411e-06,
      "loss": 0.4382,
      "step": 5089
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2670066356658936,
      "learning_rate": 1.3841289277170223e-06,
      "loss": 0.4493,
      "step": 5090
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2546443939208984,
      "learning_rate": 1.3824491713353332e-06,
      "loss": 0.4801,
      "step": 5091
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4293570518493652,
      "learning_rate": 1.3807702713066084e-06,
      "loss": 0.4665,
      "step": 5092
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.222167730331421,
      "learning_rate": 1.379092228028281e-06,
      "loss": 0.4985,
      "step": 5093
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6939810514450073,
      "learning_rate": 1.3774150418975807e-06,
      "loss": 0.494,
      "step": 5094
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.330978512763977,
      "learning_rate": 1.3757387133115346e-06,
      "loss": 0.4319,
      "step": 5095
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.118896007537842,
      "learning_rate": 1.374063242666966e-06,
      "loss": 0.4909,
      "step": 5096
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2229537963867188,
      "learning_rate": 1.3723886303604922e-06,
      "loss": 0.5226,
      "step": 5097
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3765876293182373,
      "learning_rate": 1.3707148767885348e-06,
      "loss": 0.5169,
      "step": 5098
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.429858922958374,
      "learning_rate": 1.3690419823473034e-06,
      "loss": 0.5067,
      "step": 5099
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.348372220993042,
      "learning_rate": 1.3673699474328122e-06,
      "loss": 0.5009,
      "step": 5100
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2080639600753784,
      "learning_rate": 1.3656987724408655e-06,
      "loss": 0.7032,
      "step": 5101
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3680464029312134,
      "learning_rate": 1.3640284577670692e-06,
      "loss": 0.454,
      "step": 5102
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.491603970527649,
      "learning_rate": 1.3623590038068219e-06,
      "loss": 0.4433,
      "step": 5103
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.200669527053833,
      "learning_rate": 1.3606904109553165e-06,
      "loss": 0.4257,
      "step": 5104
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0511072874069214,
      "learning_rate": 1.3590226796075518e-06,
      "loss": 0.453,
      "step": 5105
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6535866260528564,
      "learning_rate": 1.3573558101583106e-06,
      "loss": 0.4558,
      "step": 5106
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1436647176742554,
      "learning_rate": 1.3556898030021804e-06,
      "loss": 0.6665,
      "step": 5107
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.100401759147644,
      "learning_rate": 1.3540246585335403e-06,
      "loss": 0.4484,
      "step": 5108
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.2472405433654785,
      "learning_rate": 1.3523603771465648e-06,
      "loss": 0.4923,
      "step": 5109
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6060523986816406,
      "learning_rate": 1.3506969592352282e-06,
      "loss": 0.7515,
      "step": 5110
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.3957016468048096,
      "learning_rate": 1.349034405193294e-06,
      "loss": 0.4825,
      "step": 5111
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4056026935577393,
      "learning_rate": 1.3473727154143296e-06,
      "loss": 0.4949,
      "step": 5112
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4872798919677734,
      "learning_rate": 1.3457118902916883e-06,
      "loss": 0.3986,
      "step": 5113
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.390319585800171,
      "learning_rate": 1.3440519302185278e-06,
      "loss": 0.5004,
      "step": 5114
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2242182493209839,
      "learning_rate": 1.3423928355877947e-06,
      "loss": 0.4529,
      "step": 5115
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.321574091911316,
      "learning_rate": 1.3407346067922294e-06,
      "loss": 0.4094,
      "step": 5116
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2803524732589722,
      "learning_rate": 1.3390772442243776e-06,
      "loss": 0.4662,
      "step": 5117
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3647897243499756,
      "learning_rate": 1.3374207482765683e-06,
      "loss": 0.4258,
      "step": 5118
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.191253900527954,
      "learning_rate": 1.3357651193409333e-06,
      "loss": 0.6882,
      "step": 5119
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0969157218933105,
      "learning_rate": 1.3341103578093934e-06,
      "loss": 0.4672,
      "step": 5120
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.50178861618042,
      "learning_rate": 1.3324564640736664e-06,
      "loss": 0.4637,
      "step": 5121
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8519598245620728,
      "learning_rate": 1.3308034385252672e-06,
      "loss": 0.4659,
      "step": 5122
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.589455485343933,
      "learning_rate": 1.3291512815555008e-06,
      "loss": 0.4501,
      "step": 5123
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7185224294662476,
      "learning_rate": 1.32749999355547e-06,
      "loss": 0.4913,
      "step": 5124
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2000203132629395,
      "learning_rate": 1.3258495749160694e-06,
      "loss": 0.5237,
      "step": 5125
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.801529049873352,
      "learning_rate": 1.3242000260279909e-06,
      "loss": 0.5011,
      "step": 5126
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0376471281051636,
      "learning_rate": 1.3225513472817158e-06,
      "loss": 0.5089,
      "step": 5127
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.789507269859314,
      "learning_rate": 1.320903539067524e-06,
      "loss": 0.4619,
      "step": 5128
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.148067593574524,
      "learning_rate": 1.3192566017754894e-06,
      "loss": 0.4494,
      "step": 5129
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7188353538513184,
      "learning_rate": 1.3176105357954733e-06,
      "loss": 0.4632,
      "step": 5130
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1235737800598145,
      "learning_rate": 1.3159653415171402e-06,
      "loss": 0.4141,
      "step": 5131
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3897286653518677,
      "learning_rate": 1.3143210193299405e-06,
      "loss": 0.4693,
      "step": 5132
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4494173526763916,
      "learning_rate": 1.3126775696231204e-06,
      "loss": 0.4859,
      "step": 5133
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2032893896102905,
      "learning_rate": 1.311034992785723e-06,
      "loss": 0.4251,
      "step": 5134
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3702681064605713,
      "learning_rate": 1.3093932892065787e-06,
      "loss": 0.4475,
      "step": 5135
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3662532567977905,
      "learning_rate": 1.3077524592743174e-06,
      "loss": 0.4891,
      "step": 5136
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.054112434387207,
      "learning_rate": 1.3061125033773564e-06,
      "loss": 0.4688,
      "step": 5137
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.372502326965332,
      "learning_rate": 1.3044734219039108e-06,
      "loss": 0.4971,
      "step": 5138
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7491286993026733,
      "learning_rate": 1.3028352152419876e-06,
      "loss": 0.3793,
      "step": 5139
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.087489366531372,
      "learning_rate": 1.3011978837793832e-06,
      "loss": 0.4184,
      "step": 5140
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3097480535507202,
      "learning_rate": 1.2995614279036933e-06,
      "loss": 0.4908,
      "step": 5141
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5528531074523926,
      "learning_rate": 1.2979258480022983e-06,
      "loss": 0.4471,
      "step": 5142
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4678362607955933,
      "learning_rate": 1.2962911444623799e-06,
      "loss": 0.4336,
      "step": 5143
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2267463207244873,
      "learning_rate": 1.2946573176709055e-06,
      "loss": 0.4599,
      "step": 5144
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2914280891418457,
      "learning_rate": 1.293024368014636e-06,
      "loss": 0.4972,
      "step": 5145
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2014577388763428,
      "learning_rate": 1.2913922958801294e-06,
      "loss": 0.4523,
      "step": 5146
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1257227659225464,
      "learning_rate": 1.2897611016537304e-06,
      "loss": 0.4415,
      "step": 5147
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3313616514205933,
      "learning_rate": 1.2881307857215797e-06,
      "loss": 0.4293,
      "step": 5148
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.3975472450256348,
      "learning_rate": 1.286501348469606e-06,
      "loss": 0.4613,
      "step": 5149
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2255898714065552,
      "learning_rate": 1.2848727902835351e-06,
      "loss": 0.4292,
      "step": 5150
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9457247853279114,
      "learning_rate": 1.283245111548882e-06,
      "loss": 0.4186,
      "step": 5151
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.107791543006897,
      "learning_rate": 1.2816183126509519e-06,
      "loss": 0.6511,
      "step": 5152
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0656334161758423,
      "learning_rate": 1.279992393974846e-06,
      "loss": 0.3935,
      "step": 5153
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1483198404312134,
      "learning_rate": 1.2783673559054515e-06,
      "loss": 0.4712,
      "step": 5154
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6354182958602905,
      "learning_rate": 1.2767431988274538e-06,
      "loss": 0.4727,
      "step": 5155
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5054155588150024,
      "learning_rate": 1.2751199231253236e-06,
      "loss": 0.4644,
      "step": 5156
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.553381323814392,
      "learning_rate": 1.2734975291833252e-06,
      "loss": 0.4784,
      "step": 5157
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.407997488975525,
      "learning_rate": 1.271876017385517e-06,
      "loss": 0.4471,
      "step": 5158
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0466728210449219,
      "learning_rate": 1.2702553881157438e-06,
      "loss": 0.703,
      "step": 5159
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.234391212463379,
      "learning_rate": 1.2686356417576439e-06,
      "loss": 0.473,
      "step": 5160
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8142240047454834,
      "learning_rate": 1.2670167786946496e-06,
      "loss": 0.4955,
      "step": 5161
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7036758661270142,
      "learning_rate": 1.265398799309977e-06,
      "loss": 0.4412,
      "step": 5162
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8013274669647217,
      "learning_rate": 1.2637817039866402e-06,
      "loss": 0.4751,
      "step": 5163
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.0170881748199463,
      "learning_rate": 1.262165493107438e-06,
      "loss": 0.4888,
      "step": 5164
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.277538776397705,
      "learning_rate": 1.2605501670549653e-06,
      "loss": 0.5095,
      "step": 5165
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6744341850280762,
      "learning_rate": 1.2589357262116019e-06,
      "loss": 0.4825,
      "step": 5166
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1016966104507446,
      "learning_rate": 1.2573221709595246e-06,
      "loss": 0.3718,
      "step": 5167
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4773236513137817,
      "learning_rate": 1.2557095016806943e-06,
      "loss": 0.4513,
      "step": 5168
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.199238657951355,
      "learning_rate": 1.2540977187568665e-06,
      "loss": 0.5255,
      "step": 5169
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4167215824127197,
      "learning_rate": 1.2524868225695851e-06,
      "loss": 0.4607,
      "step": 5170
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.370404601097107,
      "learning_rate": 1.2508768135001808e-06,
      "loss": 0.4905,
      "step": 5171
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2135894298553467,
      "learning_rate": 1.2492676919297837e-06,
      "loss": 0.4668,
      "step": 5172
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2096397876739502,
      "learning_rate": 1.2476594582393048e-06,
      "loss": 0.5149,
      "step": 5173
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5600794553756714,
      "learning_rate": 1.2460521128094467e-06,
      "loss": 0.4231,
      "step": 5174
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1658220291137695,
      "learning_rate": 1.2444456560207058e-06,
      "loss": 0.382,
      "step": 5175
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1190600395202637,
      "learning_rate": 1.2428400882533625e-06,
      "loss": 0.7085,
      "step": 5176
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2184534072875977,
      "learning_rate": 1.2412354098874924e-06,
      "loss": 0.429,
      "step": 5177
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6534632444381714,
      "learning_rate": 1.2396316213029553e-06,
      "loss": 0.4907,
      "step": 5178
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1247432231903076,
      "learning_rate": 1.2380287228794058e-06,
      "loss": 0.4481,
      "step": 5179
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2121891975402832,
      "learning_rate": 1.236426714996281e-06,
      "loss": 0.444,
      "step": 5180
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.414454460144043,
      "learning_rate": 1.234825598032815e-06,
      "loss": 0.4287,
      "step": 5181
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9463626146316528,
      "learning_rate": 1.2332253723680231e-06,
      "loss": 0.4299,
      "step": 5182
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0559909343719482,
      "learning_rate": 1.2316260383807161e-06,
      "loss": 0.4264,
      "step": 5183
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.581770420074463,
      "learning_rate": 1.2300275964494918e-06,
      "loss": 0.4272,
      "step": 5184
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2202616930007935,
      "learning_rate": 1.228430046952735e-06,
      "loss": 0.3988,
      "step": 5185
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.16873037815094,
      "learning_rate": 1.2268333902686185e-06,
      "loss": 0.3504,
      "step": 5186
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.00873601436615,
      "learning_rate": 1.2252376267751093e-06,
      "loss": 0.4194,
      "step": 5187
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.152464747428894,
      "learning_rate": 1.223642756849956e-06,
      "loss": 0.7048,
      "step": 5188
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1540316343307495,
      "learning_rate": 1.2220487808707016e-06,
      "loss": 0.4465,
      "step": 5189
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1682920455932617,
      "learning_rate": 1.2204556992146726e-06,
      "loss": 0.4667,
      "step": 5190
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0661836862564087,
      "learning_rate": 1.2188635122589887e-06,
      "loss": 0.428,
      "step": 5191
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1806366443634033,
      "learning_rate": 1.2172722203805526e-06,
      "loss": 0.4832,
      "step": 5192
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3543082475662231,
      "learning_rate": 1.2156818239560604e-06,
      "loss": 0.5292,
      "step": 5193
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5635645389556885,
      "learning_rate": 1.21409232336199e-06,
      "loss": 0.5023,
      "step": 5194
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2174092531204224,
      "learning_rate": 1.212503718974613e-06,
      "loss": 0.5621,
      "step": 5195
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3050566911697388,
      "learning_rate": 1.2109160111699875e-06,
      "loss": 0.3881,
      "step": 5196
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.343445897102356,
      "learning_rate": 1.2093292003239565e-06,
      "loss": 0.4807,
      "step": 5197
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1587727069854736,
      "learning_rate": 1.2077432868121547e-06,
      "loss": 0.4081,
      "step": 5198
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0760446786880493,
      "learning_rate": 1.2061582710100006e-06,
      "loss": 0.4163,
      "step": 5199
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2043375968933105,
      "learning_rate": 1.2045741532927014e-06,
      "loss": 0.4631,
      "step": 5200
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.1445600986480713,
      "learning_rate": 1.2029909340352547e-06,
      "loss": 0.5192,
      "step": 5201
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.366042137145996,
      "learning_rate": 1.2014086136124398e-06,
      "loss": 0.4396,
      "step": 5202
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2837494611740112,
      "learning_rate": 1.1998271923988298e-06,
      "loss": 0.5164,
      "step": 5203
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1921879053115845,
      "learning_rate": 1.1982466707687784e-06,
      "loss": 0.5183,
      "step": 5204
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.233123540878296,
      "learning_rate": 1.1966670490964306e-06,
      "loss": 0.4262,
      "step": 5205
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2451629638671875,
      "learning_rate": 1.1950883277557185e-06,
      "loss": 0.4792,
      "step": 5206
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9966418147087097,
      "learning_rate": 1.1935105071203572e-06,
      "loss": 0.3826,
      "step": 5207
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.0658695697784424,
      "learning_rate": 1.1919335875638543e-06,
      "loss": 0.5019,
      "step": 5208
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.442794919013977,
      "learning_rate": 1.1903575694594982e-06,
      "loss": 0.4392,
      "step": 5209
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8551212549209595,
      "learning_rate": 1.1887824531803688e-06,
      "loss": 0.4956,
      "step": 5210
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.494482159614563,
      "learning_rate": 1.1872082390993294e-06,
      "loss": 0.4892,
      "step": 5211
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3777120113372803,
      "learning_rate": 1.185634927589029e-06,
      "loss": 0.4905,
      "step": 5212
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.145352840423584,
      "learning_rate": 1.184062519021908e-06,
      "loss": 0.4161,
      "step": 5213
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1590969562530518,
      "learning_rate": 1.1824910137701861e-06,
      "loss": 0.3984,
      "step": 5214
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0942548513412476,
      "learning_rate": 1.180920412205877e-06,
      "loss": 0.4122,
      "step": 5215
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0436488389968872,
      "learning_rate": 1.1793507147007716e-06,
      "loss": 0.4267,
      "step": 5216
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.826746940612793,
      "learning_rate": 1.177781921626454e-06,
      "loss": 0.3643,
      "step": 5217
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7451331615447998,
      "learning_rate": 1.1762140333542932e-06,
      "loss": 0.4975,
      "step": 5218
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1511015892028809,
      "learning_rate": 1.174647050255439e-06,
      "loss": 0.5125,
      "step": 5219
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4533586502075195,
      "learning_rate": 1.173080972700834e-06,
      "loss": 0.4763,
      "step": 5220
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1550867557525635,
      "learning_rate": 1.1715158010611987e-06,
      "loss": 0.6742,
      "step": 5221
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4239732027053833,
      "learning_rate": 1.1699515357070473e-06,
      "loss": 0.4661,
      "step": 5222
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2067009210586548,
      "learning_rate": 1.1683881770086735e-06,
      "loss": 0.4901,
      "step": 5223
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7290204763412476,
      "learning_rate": 1.166825725336157e-06,
      "loss": 0.4224,
      "step": 5224
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.497635006904602,
      "learning_rate": 1.1652641810593667e-06,
      "loss": 0.4818,
      "step": 5225
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.869317889213562,
      "learning_rate": 1.1637035445479517e-06,
      "loss": 0.4513,
      "step": 5226
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1962324380874634,
      "learning_rate": 1.1621438161713494e-06,
      "loss": 0.4979,
      "step": 5227
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4031176567077637,
      "learning_rate": 1.1605849962987836e-06,
      "loss": 0.5248,
      "step": 5228
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3667857646942139,
      "learning_rate": 1.159027085299258e-06,
      "loss": 0.4365,
      "step": 5229
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1856306791305542,
      "learning_rate": 1.157470083541566e-06,
      "loss": 0.3716,
      "step": 5230
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.914514422416687,
      "learning_rate": 1.1559139913942812e-06,
      "loss": 0.4273,
      "step": 5231
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6829432249069214,
      "learning_rate": 1.1543588092257685e-06,
      "loss": 0.4668,
      "step": 5232
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0694003105163574,
      "learning_rate": 1.152804537404169e-06,
      "loss": 0.3867,
      "step": 5233
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6268988847732544,
      "learning_rate": 1.151251176297416e-06,
      "loss": 0.4944,
      "step": 5234
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.742185354232788,
      "learning_rate": 1.1496987262732224e-06,
      "loss": 0.4624,
      "step": 5235
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3686329126358032,
      "learning_rate": 1.1481471876990858e-06,
      "loss": 0.485,
      "step": 5236
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1361842155456543,
      "learning_rate": 1.146596560942292e-06,
      "loss": 0.4666,
      "step": 5237
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6372764110565186,
      "learning_rate": 1.1450468463699033e-06,
      "loss": 0.4296,
      "step": 5238
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4072470664978027,
      "learning_rate": 1.1434980443487764e-06,
      "loss": 0.4948,
      "step": 5239
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9001047611236572,
      "learning_rate": 1.1419501552455442e-06,
      "loss": 0.438,
      "step": 5240
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3046094179153442,
      "learning_rate": 1.1404031794266246e-06,
      "loss": 0.4718,
      "step": 5241
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.793980836868286,
      "learning_rate": 1.1388571172582218e-06,
      "loss": 0.5071,
      "step": 5242
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.326981544494629,
      "learning_rate": 1.137311969106321e-06,
      "loss": 0.4209,
      "step": 5243
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.5776288509368896,
      "learning_rate": 1.1357677353366947e-06,
      "loss": 0.4972,
      "step": 5244
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2050260305404663,
      "learning_rate": 1.1342244163148935e-06,
      "loss": 0.4744,
      "step": 5245
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.469270944595337,
      "learning_rate": 1.1326820124062581e-06,
      "loss": 0.4827,
      "step": 5246
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3363481760025024,
      "learning_rate": 1.1311405239759065e-06,
      "loss": 0.4793,
      "step": 5247
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.274931788444519,
      "learning_rate": 1.1295999513887423e-06,
      "loss": 0.4473,
      "step": 5248
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1056151390075684,
      "learning_rate": 1.1280602950094532e-06,
      "loss": 0.4532,
      "step": 5249
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.9260857105255127,
      "learning_rate": 1.12652155520251e-06,
      "loss": 0.5094,
      "step": 5250
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.4739391803741455,
      "learning_rate": 1.1249837323321666e-06,
      "loss": 0.5073,
      "step": 5251
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3294532299041748,
      "learning_rate": 1.1234468267624581e-06,
      "loss": 0.4922,
      "step": 5252
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5017918348312378,
      "learning_rate": 1.121910838857202e-06,
      "loss": 0.4044,
      "step": 5253
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.771401047706604,
      "learning_rate": 1.1203757689800026e-06,
      "loss": 0.4776,
      "step": 5254
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.334784746170044,
      "learning_rate": 1.1188416174942423e-06,
      "loss": 0.4538,
      "step": 5255
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.779727458953857,
      "learning_rate": 1.11730838476309e-06,
      "loss": 0.4769,
      "step": 5256
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4938327074050903,
      "learning_rate": 1.115776071149493e-06,
      "loss": 0.4967,
      "step": 5257
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.484053373336792,
      "learning_rate": 1.1142446770161858e-06,
      "loss": 0.5462,
      "step": 5258
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5999958515167236,
      "learning_rate": 1.1127142027256809e-06,
      "loss": 0.4306,
      "step": 5259
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1755390167236328,
      "learning_rate": 1.1111846486402733e-06,
      "loss": 0.6563,
      "step": 5260
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2558156251907349,
      "learning_rate": 1.1096560151220464e-06,
      "loss": 0.4738,
      "step": 5261
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4738284349441528,
      "learning_rate": 1.108128302532857e-06,
      "loss": 0.5387,
      "step": 5262
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1370048522949219,
      "learning_rate": 1.1066015112343503e-06,
      "loss": 0.5039,
      "step": 5263
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.125353455543518,
      "learning_rate": 1.1050756415879504e-06,
      "loss": 0.4425,
      "step": 5264
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3548436164855957,
      "learning_rate": 1.1035506939548618e-06,
      "loss": 0.405,
      "step": 5265
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4237253665924072,
      "learning_rate": 1.1020266686960762e-06,
      "loss": 0.5021,
      "step": 5266
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5300257205963135,
      "learning_rate": 1.1005035661723595e-06,
      "loss": 0.4091,
      "step": 5267
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8265262842178345,
      "learning_rate": 1.0989813867442677e-06,
      "loss": 0.4812,
      "step": 5268
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3431875705718994,
      "learning_rate": 1.0974601307721295e-06,
      "loss": 0.4224,
      "step": 5269
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2974804639816284,
      "learning_rate": 1.095939798616063e-06,
      "loss": 0.4085,
      "step": 5270
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.268454670906067,
      "learning_rate": 1.0944203906359607e-06,
      "loss": 0.4601,
      "step": 5271
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3931277990341187,
      "learning_rate": 1.092901907191501e-06,
      "loss": 0.449,
      "step": 5272
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2744719982147217,
      "learning_rate": 1.0913843486421427e-06,
      "loss": 0.5182,
      "step": 5273
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6259069442749023,
      "learning_rate": 1.0898677153471239e-06,
      "loss": 0.4165,
      "step": 5274
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7017569541931152,
      "learning_rate": 1.0883520076654652e-06,
      "loss": 0.3705,
      "step": 5275
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2849503755569458,
      "learning_rate": 1.086837225955968e-06,
      "loss": 0.4282,
      "step": 5276
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8516933917999268,
      "learning_rate": 1.085323370577212e-06,
      "loss": 0.4724,
      "step": 5277
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1455886363983154,
      "learning_rate": 1.0838104418875622e-06,
      "loss": 0.4514,
      "step": 5278
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.558569312095642,
      "learning_rate": 1.0822984402451591e-06,
      "loss": 0.4329,
      "step": 5279
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1546862125396729,
      "learning_rate": 1.0807873660079293e-06,
      "loss": 0.4282,
      "step": 5280
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1991701126098633,
      "learning_rate": 1.0792772195335749e-06,
      "loss": 0.4415,
      "step": 5281
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5490829944610596,
      "learning_rate": 1.0777680011795815e-06,
      "loss": 0.5113,
      "step": 5282
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1298505067825317,
      "learning_rate": 1.0762597113032124e-06,
      "loss": 0.4132,
      "step": 5283
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6462957859039307,
      "learning_rate": 1.0747523502615136e-06,
      "loss": 0.4495,
      "step": 5284
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.578015089035034,
      "learning_rate": 1.073245918411312e-06,
      "loss": 0.4241,
      "step": 5285
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2173229455947876,
      "learning_rate": 1.0717404161092098e-06,
      "loss": 0.4568,
      "step": 5286
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.709384799003601,
      "learning_rate": 1.0702358437115945e-06,
      "loss": 0.417,
      "step": 5287
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3775248527526855,
      "learning_rate": 1.0687322015746304e-06,
      "loss": 0.4984,
      "step": 5288
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3653825521469116,
      "learning_rate": 1.06722949005426e-06,
      "loss": 0.503,
      "step": 5289
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.197861671447754,
      "learning_rate": 1.065727709506212e-06,
      "loss": 0.4184,
      "step": 5290
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1938765048980713,
      "learning_rate": 1.0642268602859862e-06,
      "loss": 0.5076,
      "step": 5291
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3841642141342163,
      "learning_rate": 1.0627269427488702e-06,
      "loss": 0.4899,
      "step": 5292
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.479785680770874,
      "learning_rate": 1.0612279572499234e-06,
      "loss": 0.4385,
      "step": 5293
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4520254135131836,
      "learning_rate": 1.0597299041439901e-06,
      "loss": 0.4482,
      "step": 5294
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5893441438674927,
      "learning_rate": 1.0582327837856937e-06,
      "loss": 0.468,
      "step": 5295
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.256061315536499,
      "learning_rate": 1.0567365965294318e-06,
      "loss": 0.4416,
      "step": 5296
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.26748788356781,
      "learning_rate": 1.0552413427293885e-06,
      "loss": 0.4527,
      "step": 5297
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.202327847480774,
      "learning_rate": 1.053747022739518e-06,
      "loss": 0.4763,
      "step": 5298
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5768507719039917,
      "learning_rate": 1.0522536369135634e-06,
      "loss": 0.5034,
      "step": 5299
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.164113998413086,
      "learning_rate": 1.050761185605038e-06,
      "loss": 0.4762,
      "step": 5300
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3254849910736084,
      "learning_rate": 1.0492696691672382e-06,
      "loss": 0.465,
      "step": 5301
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.3539252281188965,
      "learning_rate": 1.0477790879532397e-06,
      "loss": 0.4579,
      "step": 5302
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7476279735565186,
      "learning_rate": 1.0462894423158936e-06,
      "loss": 0.5118,
      "step": 5303
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6620429754257202,
      "learning_rate": 1.0448007326078336e-06,
      "loss": 0.4645,
      "step": 5304
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.064635753631592,
      "learning_rate": 1.0433129591814668e-06,
      "loss": 0.5245,
      "step": 5305
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1933202743530273,
      "learning_rate": 1.0418261223889836e-06,
      "loss": 0.4364,
      "step": 5306
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.386518955230713,
      "learning_rate": 1.0403402225823517e-06,
      "loss": 0.4613,
      "step": 5307
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5397158861160278,
      "learning_rate": 1.038855260113313e-06,
      "loss": 0.4771,
      "step": 5308
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.6200342178344727,
      "learning_rate": 1.0373712353333925e-06,
      "loss": 0.4927,
      "step": 5309
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.2491869926452637,
      "learning_rate": 1.0358881485938898e-06,
      "loss": 0.4164,
      "step": 5310
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.354323148727417,
      "learning_rate": 1.0344060002458862e-06,
      "loss": 0.6949,
      "step": 5311
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4152737855911255,
      "learning_rate": 1.0329247906402345e-06,
      "loss": 0.4369,
      "step": 5312
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2759592533111572,
      "learning_rate": 1.0314445201275737e-06,
      "loss": 0.4454,
      "step": 5313
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.144474744796753,
      "learning_rate": 1.0299651890583128e-06,
      "loss": 0.4829,
      "step": 5314
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2991188764572144,
      "learning_rate": 1.0284867977826418e-06,
      "loss": 0.5267,
      "step": 5315
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.232662320137024,
      "learning_rate": 1.0270093466505292e-06,
      "loss": 0.5004,
      "step": 5316
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.456958532333374,
      "learning_rate": 1.0255328360117207e-06,
      "loss": 0.4416,
      "step": 5317
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1746290922164917,
      "learning_rate": 1.0240572662157355e-06,
      "loss": 0.4865,
      "step": 5318
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3923572301864624,
      "learning_rate": 1.0225826376118769e-06,
      "loss": 0.4398,
      "step": 5319
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4473084211349487,
      "learning_rate": 1.0211089505492178e-06,
      "loss": 0.4819,
      "step": 5320
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2220197916030884,
      "learning_rate": 1.019636205376615e-06,
      "loss": 0.4158,
      "step": 5321
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.254901647567749,
      "learning_rate": 1.0181644024426968e-06,
      "loss": 0.4529,
      "step": 5322
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3547804355621338,
      "learning_rate": 1.0166935420958736e-06,
      "loss": 0.4768,
      "step": 5323
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3289376497268677,
      "learning_rate": 1.0152236246843272e-06,
      "loss": 0.4898,
      "step": 5324
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1931613683700562,
      "learning_rate": 1.013754650556022e-06,
      "loss": 0.4701,
      "step": 5325
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.626822590827942,
      "learning_rate": 1.0122866200586945e-06,
      "loss": 0.4727,
      "step": 5326
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4110437631607056,
      "learning_rate": 1.0108195335398568e-06,
      "loss": 0.4807,
      "step": 5327
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3182185888290405,
      "learning_rate": 1.0093533913468056e-06,
      "loss": 0.4289,
      "step": 5328
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.489305853843689,
      "learning_rate": 1.007888193826606e-06,
      "loss": 0.4747,
      "step": 5329
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.192521095275879,
      "learning_rate": 1.0064239413261006e-06,
      "loss": 0.4594,
      "step": 5330
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.406035304069519,
      "learning_rate": 1.0049606341919127e-06,
      "loss": 0.4373,
      "step": 5331
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1903780698776245,
      "learning_rate": 1.0034982727704363e-06,
      "loss": 0.4112,
      "step": 5332
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.207593321800232,
      "learning_rate": 1.0020368574078464e-06,
      "loss": 0.4821,
      "step": 5333
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.392045021057129,
      "learning_rate": 1.0005763884500901e-06,
      "loss": 0.425,
      "step": 5334
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4266036748886108,
      "learning_rate": 9.99116866242894e-07,
      "loss": 0.4555,
      "step": 5335
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8962862491607666,
      "learning_rate": 9.976582911317562e-07,
      "loss": 0.4893,
      "step": 5336
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2281192541122437,
      "learning_rate": 9.96200663461956e-07,
      "loss": 0.4434,
      "step": 5337
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.287347674369812,
      "learning_rate": 9.947439835785421e-07,
      "loss": 0.4907,
      "step": 5338
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4154276847839355,
      "learning_rate": 9.932882518263454e-07,
      "loss": 0.4553,
      "step": 5339
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2857893705368042,
      "learning_rate": 9.918334685499686e-07,
      "loss": 0.4096,
      "step": 5340
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.158995270729065,
      "learning_rate": 9.903796340937893e-07,
      "loss": 0.6343,
      "step": 5341
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5595191717147827,
      "learning_rate": 9.889267488019645e-07,
      "loss": 0.4836,
      "step": 5342
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4988987445831299,
      "learning_rate": 9.87474813018421e-07,
      "loss": 0.4284,
      "step": 5343
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1118453741073608,
      "learning_rate": 9.860238270868633e-07,
      "loss": 0.4692,
      "step": 5344
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3717315196990967,
      "learning_rate": 9.84573791350774e-07,
      "loss": 0.4932,
      "step": 5345
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.410163402557373,
      "learning_rate": 9.831247061534049e-07,
      "loss": 0.459,
      "step": 5346
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1952134370803833,
      "learning_rate": 9.816765718377896e-07,
      "loss": 0.3666,
      "step": 5347
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4420783519744873,
      "learning_rate": 9.80229388746729e-07,
      "loss": 0.4253,
      "step": 5348
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.27654230594635,
      "learning_rate": 9.78783157222805e-07,
      "loss": 0.4723,
      "step": 5349
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4250271320343018,
      "learning_rate": 9.773378776083736e-07,
      "loss": 0.4637,
      "step": 5350
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3461068868637085,
      "learning_rate": 9.758935502455607e-07,
      "loss": 0.5024,
      "step": 5351
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4528136253356934,
      "learning_rate": 9.74450175476273e-07,
      "loss": 0.3798,
      "step": 5352
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3247836828231812,
      "learning_rate": 9.730077536421862e-07,
      "loss": 0.4098,
      "step": 5353
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4236869812011719,
      "learning_rate": 9.715662850847547e-07,
      "loss": 0.466,
      "step": 5354
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2377299070358276,
      "learning_rate": 9.701257701452049e-07,
      "loss": 0.3864,
      "step": 5355
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.3298370838165283,
      "learning_rate": 9.686862091645366e-07,
      "loss": 0.4519,
      "step": 5356
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2358368635177612,
      "learning_rate": 9.672476024835276e-07,
      "loss": 0.427,
      "step": 5357
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2655580043792725,
      "learning_rate": 9.658099504427249e-07,
      "loss": 0.4684,
      "step": 5358
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3600386381149292,
      "learning_rate": 9.643732533824545e-07,
      "loss": 0.4797,
      "step": 5359
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2506697177886963,
      "learning_rate": 9.62937511642812e-07,
      "loss": 0.4917,
      "step": 5360
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2352408170700073,
      "learning_rate": 9.61502725563669e-07,
      "loss": 0.4617,
      "step": 5361
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2039283514022827,
      "learning_rate": 9.60068895484672e-07,
      "loss": 0.3967,
      "step": 5362
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3075278997421265,
      "learning_rate": 9.58636021745238e-07,
      "loss": 0.4667,
      "step": 5363
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1934106349945068,
      "learning_rate": 9.572041046845604e-07,
      "loss": 0.4624,
      "step": 5364
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5741909742355347,
      "learning_rate": 9.557731446416046e-07,
      "loss": 0.3951,
      "step": 5365
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2925388813018799,
      "learning_rate": 9.543431419551108e-07,
      "loss": 0.4686,
      "step": 5366
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2003880739212036,
      "learning_rate": 9.529140969635914e-07,
      "loss": 0.4452,
      "step": 5367
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.51370108127594,
      "learning_rate": 9.514860100053308e-07,
      "loss": 0.4754,
      "step": 5368
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.807694911956787,
      "learning_rate": 9.500588814183909e-07,
      "loss": 0.5091,
      "step": 5369
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0238769054412842,
      "learning_rate": 9.486327115406013e-07,
      "loss": 0.4178,
      "step": 5370
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1914355754852295,
      "learning_rate": 9.472075007095704e-07,
      "loss": 0.4781,
      "step": 5371
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3640578985214233,
      "learning_rate": 9.457832492626739e-07,
      "loss": 0.4406,
      "step": 5372
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3747899532318115,
      "learning_rate": 9.443599575370638e-07,
      "loss": 0.4465,
      "step": 5373
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.329126000404358,
      "learning_rate": 9.429376258696665e-07,
      "loss": 0.4938,
      "step": 5374
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0654733180999756,
      "learning_rate": 9.415162545971757e-07,
      "loss": 0.415,
      "step": 5375
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9997240304946899,
      "learning_rate": 9.400958440560632e-07,
      "loss": 0.3887,
      "step": 5376
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3998608589172363,
      "learning_rate": 9.386763945825689e-07,
      "loss": 0.4684,
      "step": 5377
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4656070470809937,
      "learning_rate": 9.372579065127102e-07,
      "loss": 0.5073,
      "step": 5378
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5117111206054688,
      "learning_rate": 9.358403801822724e-07,
      "loss": 0.5278,
      "step": 5379
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0668411254882812,
      "learning_rate": 9.344238159268132e-07,
      "loss": 0.6958,
      "step": 5380
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8733152151107788,
      "learning_rate": 9.330082140816677e-07,
      "loss": 0.4077,
      "step": 5381
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5342447757720947,
      "learning_rate": 9.315935749819361e-07,
      "loss": 0.4482,
      "step": 5382
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7789664268493652,
      "learning_rate": 9.301798989624961e-07,
      "loss": 0.437,
      "step": 5383
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1473517417907715,
      "learning_rate": 9.287671863579967e-07,
      "loss": 0.3979,
      "step": 5384
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1428507566452026,
      "learning_rate": 9.273554375028548e-07,
      "loss": 0.3735,
      "step": 5385
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3112701177597046,
      "learning_rate": 9.259446527312654e-07,
      "loss": 0.4525,
      "step": 5386
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4192365407943726,
      "learning_rate": 9.245348323771886e-07,
      "loss": 0.4834,
      "step": 5387
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5864051580429077,
      "learning_rate": 9.231259767743622e-07,
      "loss": 0.4944,
      "step": 5388
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1974706649780273,
      "learning_rate": 9.217180862562914e-07,
      "loss": 0.4976,
      "step": 5389
    },
    {
      "epoch": 0.81,
      "grad_norm": 8.249112129211426,
      "learning_rate": 9.203111611562554e-07,
      "loss": 0.4296,
      "step": 5390
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5959347486495972,
      "learning_rate": 9.189052018073042e-07,
      "loss": 0.379,
      "step": 5391
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1972661018371582,
      "learning_rate": 9.175002085422569e-07,
      "loss": 0.4623,
      "step": 5392
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5062038898468018,
      "learning_rate": 9.160961816937086e-07,
      "loss": 0.4353,
      "step": 5393
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0668662786483765,
      "learning_rate": 9.146931215940197e-07,
      "loss": 0.5439,
      "step": 5394
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5103248357772827,
      "learning_rate": 9.132910285753294e-07,
      "loss": 0.4555,
      "step": 5395
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1764910221099854,
      "learning_rate": 9.11889902969541e-07,
      "loss": 0.4441,
      "step": 5396
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3507524728775024,
      "learning_rate": 9.104897451083305e-07,
      "loss": 0.4,
      "step": 5397
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.606834053993225,
      "learning_rate": 9.090905553231488e-07,
      "loss": 0.4794,
      "step": 5398
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8384846448898315,
      "learning_rate": 9.076923339452115e-07,
      "loss": 0.511,
      "step": 5399
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1683340072631836,
      "learning_rate": 9.062950813055099e-07,
      "loss": 0.4209,
      "step": 5400
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8371180295944214,
      "learning_rate": 9.048987977348029e-07,
      "loss": 0.4438,
      "step": 5401
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.465186595916748,
      "learning_rate": 9.035034835636225e-07,
      "loss": 0.461,
      "step": 5402
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1410506963729858,
      "learning_rate": 9.021091391222691e-07,
      "loss": 0.417,
      "step": 5403
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.1951990127563477,
      "learning_rate": 9.007157647408132e-07,
      "loss": 0.4408,
      "step": 5404
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2945020198822021,
      "learning_rate": 8.993233607490981e-07,
      "loss": 0.4713,
      "step": 5405
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2793949842453003,
      "learning_rate": 8.979319274767362e-07,
      "loss": 0.781,
      "step": 5406
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4018750190734863,
      "learning_rate": 8.965414652531112e-07,
      "loss": 0.4674,
      "step": 5407
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1615748405456543,
      "learning_rate": 8.951519744073744e-07,
      "loss": 0.4626,
      "step": 5408
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5405861139297485,
      "learning_rate": 8.937634552684471e-07,
      "loss": 0.4866,
      "step": 5409
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.375238299369812,
      "learning_rate": 8.923759081650247e-07,
      "loss": 0.5469,
      "step": 5410
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.172909140586853,
      "learning_rate": 8.90989333425567e-07,
      "loss": 0.4085,
      "step": 5411
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.168815016746521,
      "learning_rate": 8.896037313783096e-07,
      "loss": 0.4269,
      "step": 5412
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2489066123962402,
      "learning_rate": 8.88219102351251e-07,
      "loss": 0.489,
      "step": 5413
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.585219383239746,
      "learning_rate": 8.868354466721668e-07,
      "loss": 0.4045,
      "step": 5414
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2475965023040771,
      "learning_rate": 8.854527646685962e-07,
      "loss": 0.4421,
      "step": 5415
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.293412685394287,
      "learning_rate": 8.840710566678479e-07,
      "loss": 0.5252,
      "step": 5416
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2631418704986572,
      "learning_rate": 8.826903229970074e-07,
      "loss": 0.4961,
      "step": 5417
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1385117769241333,
      "learning_rate": 8.813105639829206e-07,
      "loss": 0.4247,
      "step": 5418
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1013306379318237,
      "learning_rate": 8.799317799522089e-07,
      "loss": 0.4216,
      "step": 5419
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.653403878211975,
      "learning_rate": 8.785539712312591e-07,
      "loss": 0.4793,
      "step": 5420
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1705212593078613,
      "learning_rate": 8.771771381462274e-07,
      "loss": 0.7433,
      "step": 5421
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2620055675506592,
      "learning_rate": 8.758012810230426e-07,
      "loss": 0.4702,
      "step": 5422
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3503426313400269,
      "learning_rate": 8.744264001873976e-07,
      "loss": 0.4309,
      "step": 5423
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4795517921447754,
      "learning_rate": 8.730524959647585e-07,
      "loss": 0.4546,
      "step": 5424
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2181427478790283,
      "learning_rate": 8.716795686803564e-07,
      "loss": 0.4232,
      "step": 5425
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8508806228637695,
      "learning_rate": 8.703076186591958e-07,
      "loss": 0.4388,
      "step": 5426
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.081610918045044,
      "learning_rate": 8.68936646226044e-07,
      "loss": 0.5278,
      "step": 5427
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2297916412353516,
      "learning_rate": 8.675666517054416e-07,
      "loss": 0.4899,
      "step": 5428
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.176468849182129,
      "learning_rate": 8.661976354216972e-07,
      "loss": 0.6996,
      "step": 5429
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3014782667160034,
      "learning_rate": 8.648295976988846e-07,
      "loss": 0.4664,
      "step": 5430
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0820448398590088,
      "learning_rate": 8.634625388608498e-07,
      "loss": 0.419,
      "step": 5431
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6382142305374146,
      "learning_rate": 8.620964592312048e-07,
      "loss": 0.478,
      "step": 5432
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.841234803199768,
      "learning_rate": 8.607313591333288e-07,
      "loss": 0.4502,
      "step": 5433
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5956393480300903,
      "learning_rate": 8.593672388903734e-07,
      "loss": 0.4978,
      "step": 5434
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2054696083068848,
      "learning_rate": 8.58004098825253e-07,
      "loss": 0.4896,
      "step": 5435
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.093029260635376,
      "learning_rate": 8.566419392606546e-07,
      "loss": 0.4325,
      "step": 5436
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0334421396255493,
      "learning_rate": 8.552807605190288e-07,
      "loss": 0.4552,
      "step": 5437
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2544032335281372,
      "learning_rate": 8.539205629225977e-07,
      "loss": 0.4633,
      "step": 5438
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.584778904914856,
      "learning_rate": 8.525613467933508e-07,
      "loss": 0.459,
      "step": 5439
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.409818172454834,
      "learning_rate": 8.512031124530412e-07,
      "loss": 0.4767,
      "step": 5440
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.117802381515503,
      "learning_rate": 8.498458602231957e-07,
      "loss": 0.4842,
      "step": 5441
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.363166093826294,
      "learning_rate": 8.484895904251023e-07,
      "loss": 0.4722,
      "step": 5442
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.496674060821533,
      "learning_rate": 8.471343033798224e-07,
      "loss": 0.4575,
      "step": 5443
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2398420572280884,
      "learning_rate": 8.457799994081806e-07,
      "loss": 0.4816,
      "step": 5444
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3220828771591187,
      "learning_rate": 8.444266788307687e-07,
      "loss": 0.4376,
      "step": 5445
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4883257150650024,
      "learning_rate": 8.430743419679494e-07,
      "loss": 0.5358,
      "step": 5446
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2540591955184937,
      "learning_rate": 8.417229891398487e-07,
      "loss": 0.5105,
      "step": 5447
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0343929529190063,
      "learning_rate": 8.403726206663626e-07,
      "loss": 0.4279,
      "step": 5448
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3430250883102417,
      "learning_rate": 8.390232368671508e-07,
      "loss": 0.4441,
      "step": 5449
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6662793159484863,
      "learning_rate": 8.376748380616434e-07,
      "loss": 0.4599,
      "step": 5450
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.284920573234558,
      "learning_rate": 8.363274245690362e-07,
      "loss": 0.4496,
      "step": 5451
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9080371856689453,
      "learning_rate": 8.349809967082889e-07,
      "loss": 0.4548,
      "step": 5452
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5381861925125122,
      "learning_rate": 8.336355547981328e-07,
      "loss": 0.5605,
      "step": 5453
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5508874654769897,
      "learning_rate": 8.322910991570616e-07,
      "loss": 0.4431,
      "step": 5454
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3386305570602417,
      "learning_rate": 8.309476301033387e-07,
      "loss": 0.4537,
      "step": 5455
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9362632036209106,
      "learning_rate": 8.296051479549899e-07,
      "loss": 0.5259,
      "step": 5456
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3521007299423218,
      "learning_rate": 8.282636530298133e-07,
      "loss": 0.501,
      "step": 5457
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6749563217163086,
      "learning_rate": 8.26923145645368e-07,
      "loss": 0.3772,
      "step": 5458
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.053275227546692,
      "learning_rate": 8.255836261189798e-07,
      "loss": 0.3974,
      "step": 5459
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5748459100723267,
      "learning_rate": 8.242450947677455e-07,
      "loss": 0.4038,
      "step": 5460
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2901804447174072,
      "learning_rate": 8.229075519085211e-07,
      "loss": 0.4779,
      "step": 5461
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3697896003723145,
      "learning_rate": 8.21570997857934e-07,
      "loss": 0.4783,
      "step": 5462
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9820642471313477,
      "learning_rate": 8.202354329323769e-07,
      "loss": 0.4807,
      "step": 5463
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2954175472259521,
      "learning_rate": 8.189008574480039e-07,
      "loss": 0.4646,
      "step": 5464
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4634034633636475,
      "learning_rate": 8.175672717207412e-07,
      "loss": 0.5234,
      "step": 5465
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2325440645217896,
      "learning_rate": 8.162346760662748e-07,
      "loss": 0.4727,
      "step": 5466
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2394028902053833,
      "learning_rate": 8.149030708000616e-07,
      "loss": 0.5044,
      "step": 5467
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3381187915802002,
      "learning_rate": 8.135724562373193e-07,
      "loss": 0.4404,
      "step": 5468
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.296706199645996,
      "learning_rate": 8.122428326930348e-07,
      "loss": 0.4432,
      "step": 5469
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2845309972763062,
      "learning_rate": 8.109142004819592e-07,
      "loss": 0.4216,
      "step": 5470
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4906898736953735,
      "learning_rate": 8.09586559918606e-07,
      "loss": 0.4749,
      "step": 5471
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5548452138900757,
      "learning_rate": 8.082599113172584e-07,
      "loss": 0.398,
      "step": 5472
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0641905069351196,
      "learning_rate": 8.069342549919645e-07,
      "loss": 0.4778,
      "step": 5473
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9879810810089111,
      "learning_rate": 8.056095912565337e-07,
      "loss": 0.4986,
      "step": 5474
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2194430828094482,
      "learning_rate": 8.04285920424544e-07,
      "loss": 0.489,
      "step": 5475
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3343204259872437,
      "learning_rate": 8.029632428093359e-07,
      "loss": 0.4534,
      "step": 5476
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0676742792129517,
      "learning_rate": 8.016415587240178e-07,
      "loss": 0.6859,
      "step": 5477
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4011814594268799,
      "learning_rate": 8.003208684814584e-07,
      "loss": 0.3939,
      "step": 5478
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9994133710861206,
      "learning_rate": 7.99001172394297e-07,
      "loss": 0.498,
      "step": 5479
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.203988790512085,
      "learning_rate": 7.976824707749309e-07,
      "loss": 0.7255,
      "step": 5480
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.575416922569275,
      "learning_rate": 7.963647639355287e-07,
      "loss": 0.452,
      "step": 5481
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7622084617614746,
      "learning_rate": 7.950480521880194e-07,
      "loss": 0.4658,
      "step": 5482
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9041025638580322,
      "learning_rate": 7.937323358440935e-07,
      "loss": 0.4144,
      "step": 5483
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.4974524974823,
      "learning_rate": 7.924176152152158e-07,
      "loss": 0.3934,
      "step": 5484
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1501387357711792,
      "learning_rate": 7.911038906126051e-07,
      "loss": 0.467,
      "step": 5485
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.132079839706421,
      "learning_rate": 7.89791162347251e-07,
      "loss": 0.4816,
      "step": 5486
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1370186805725098,
      "learning_rate": 7.884794307299037e-07,
      "loss": 0.4807,
      "step": 5487
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5066311359405518,
      "learning_rate": 7.871686960710773e-07,
      "loss": 0.4515,
      "step": 5488
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3449351787567139,
      "learning_rate": 7.858589586810539e-07,
      "loss": 0.4116,
      "step": 5489
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.593930959701538,
      "learning_rate": 7.845502188698745e-07,
      "loss": 0.4513,
      "step": 5490
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9396947622299194,
      "learning_rate": 7.832424769473479e-07,
      "loss": 0.3981,
      "step": 5491
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.262679934501648,
      "learning_rate": 7.819357332230437e-07,
      "loss": 0.4111,
      "step": 5492
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1136634349822998,
      "learning_rate": 7.806299880062978e-07,
      "loss": 0.4473,
      "step": 5493
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3260810375213623,
      "learning_rate": 7.793252416062075e-07,
      "loss": 0.4802,
      "step": 5494
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.086707353591919,
      "learning_rate": 7.780214943316344e-07,
      "loss": 0.4218,
      "step": 5495
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8305227756500244,
      "learning_rate": 7.767187464912063e-07,
      "loss": 0.4473,
      "step": 5496
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9984086751937866,
      "learning_rate": 7.754169983933085e-07,
      "loss": 0.4904,
      "step": 5497
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.08143949508667,
      "learning_rate": 7.741162503460959e-07,
      "loss": 0.6941,
      "step": 5498
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4042253494262695,
      "learning_rate": 7.728165026574829e-07,
      "loss": 0.4834,
      "step": 5499
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3345669507980347,
      "learning_rate": 7.715177556351467e-07,
      "loss": 0.4888,
      "step": 5500
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7057400941848755,
      "learning_rate": 7.702200095865315e-07,
      "loss": 0.433,
      "step": 5501
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.504020094871521,
      "learning_rate": 7.689232648188394e-07,
      "loss": 0.4751,
      "step": 5502
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.312827467918396,
      "learning_rate": 7.676275216390406e-07,
      "loss": 0.4571,
      "step": 5503
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2724573612213135,
      "learning_rate": 7.663327803538628e-07,
      "loss": 0.4953,
      "step": 5504
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0553594827651978,
      "learning_rate": 7.650390412698011e-07,
      "loss": 0.4331,
      "step": 5505
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0730230808258057,
      "learning_rate": 7.637463046931132e-07,
      "loss": 0.7009,
      "step": 5506
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.365461826324463,
      "learning_rate": 7.624545709298153e-07,
      "loss": 0.4449,
      "step": 5507
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7764140367507935,
      "learning_rate": 7.611638402856908e-07,
      "loss": 0.4723,
      "step": 5508
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0977987051010132,
      "learning_rate": 7.598741130662817e-07,
      "loss": 0.6406,
      "step": 5509
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.513230800628662,
      "learning_rate": 7.585853895768974e-07,
      "loss": 0.4592,
      "step": 5510
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0503628253936768,
      "learning_rate": 7.572976701226043e-07,
      "loss": 0.4419,
      "step": 5511
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.1450631618499756,
      "learning_rate": 7.56010955008234e-07,
      "loss": 0.4811,
      "step": 5512
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2651246786117554,
      "learning_rate": 7.547252445383807e-07,
      "loss": 0.458,
      "step": 5513
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6284289360046387,
      "learning_rate": 7.534405390173994e-07,
      "loss": 0.5002,
      "step": 5514
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2343732118606567,
      "learning_rate": 7.521568387494083e-07,
      "loss": 0.5318,
      "step": 5515
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2637594938278198,
      "learning_rate": 7.508741440382866e-07,
      "loss": 0.4734,
      "step": 5516
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4760719537734985,
      "learning_rate": 7.49592455187676e-07,
      "loss": 0.5104,
      "step": 5517
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4118647575378418,
      "learning_rate": 7.483117725009814e-07,
      "loss": 0.4547,
      "step": 5518
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9403373003005981,
      "learning_rate": 7.470320962813665e-07,
      "loss": 0.4447,
      "step": 5519
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2282445430755615,
      "learning_rate": 7.457534268317607e-07,
      "loss": 0.4696,
      "step": 5520
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3471035957336426,
      "learning_rate": 7.444757644548505e-07,
      "loss": 0.5085,
      "step": 5521
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1865475177764893,
      "learning_rate": 7.431991094530883e-07,
      "loss": 0.4214,
      "step": 5522
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.0464394092559814,
      "learning_rate": 7.419234621286853e-07,
      "loss": 0.4505,
      "step": 5523
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4716105461120605,
      "learning_rate": 7.406488227836139e-07,
      "loss": 0.4922,
      "step": 5524
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4202826023101807,
      "learning_rate": 7.393751917196112e-07,
      "loss": 0.4797,
      "step": 5525
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1371363401412964,
      "learning_rate": 7.381025692381705e-07,
      "loss": 0.4324,
      "step": 5526
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1017483472824097,
      "learning_rate": 7.368309556405523e-07,
      "loss": 0.3848,
      "step": 5527
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.63853120803833,
      "learning_rate": 7.355603512277726e-07,
      "loss": 0.4595,
      "step": 5528
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4073748588562012,
      "learning_rate": 7.342907563006125e-07,
      "loss": 0.3975,
      "step": 5529
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6277443170547485,
      "learning_rate": 7.330221711596136e-07,
      "loss": 0.4894,
      "step": 5530
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2105830907821655,
      "learning_rate": 7.317545961050748e-07,
      "loss": 0.4711,
      "step": 5531
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.319808006286621,
      "learning_rate": 7.304880314370621e-07,
      "loss": 0.5031,
      "step": 5532
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3166050910949707,
      "learning_rate": 7.292224774553958e-07,
      "loss": 0.4961,
      "step": 5533
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1974438428878784,
      "learning_rate": 7.279579344596627e-07,
      "loss": 0.4546,
      "step": 5534
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.209348201751709,
      "learning_rate": 7.266944027492062e-07,
      "loss": 0.4886,
      "step": 5535
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.099717140197754,
      "learning_rate": 7.254318826231305e-07,
      "loss": 0.4885,
      "step": 5536
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1217186450958252,
      "learning_rate": 7.241703743803046e-07,
      "loss": 0.503,
      "step": 5537
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.320224642753601,
      "learning_rate": 7.229098783193517e-07,
      "loss": 0.4139,
      "step": 5538
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5448534488677979,
      "learning_rate": 7.216503947386605e-07,
      "loss": 0.3961,
      "step": 5539
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6031211614608765,
      "learning_rate": 7.203919239363794e-07,
      "loss": 0.4565,
      "step": 5540
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5772968530654907,
      "learning_rate": 7.191344662104127e-07,
      "loss": 0.4348,
      "step": 5541
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3457868099212646,
      "learning_rate": 7.178780218584308e-07,
      "loss": 0.5283,
      "step": 5542
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2546379566192627,
      "learning_rate": 7.166225911778591e-07,
      "loss": 0.4783,
      "step": 5543
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.333397388458252,
      "learning_rate": 7.153681744658874e-07,
      "loss": 0.3611,
      "step": 5544
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4867981672286987,
      "learning_rate": 7.141147720194619e-07,
      "loss": 0.53,
      "step": 5545
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.18803071975708,
      "learning_rate": 7.128623841352916e-07,
      "loss": 0.4914,
      "step": 5546
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.393268585205078,
      "learning_rate": 7.116110111098435e-07,
      "loss": 0.4794,
      "step": 5547
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.397573709487915,
      "learning_rate": 7.103606532393431e-07,
      "loss": 0.4149,
      "step": 5548
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.001762866973877,
      "learning_rate": 7.091113108197794e-07,
      "loss": 0.4463,
      "step": 5549
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7267104387283325,
      "learning_rate": 7.078629841468965e-07,
      "loss": 0.5125,
      "step": 5550
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.77421236038208,
      "learning_rate": 7.066156735162044e-07,
      "loss": 0.5101,
      "step": 5551
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1714750528335571,
      "learning_rate": 7.05369379222966e-07,
      "loss": 0.4312,
      "step": 5552
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4417266845703125,
      "learning_rate": 7.041241015622063e-07,
      "loss": 0.4882,
      "step": 5553
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2567822933197021,
      "learning_rate": 7.028798408287108e-07,
      "loss": 0.4307,
      "step": 5554
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1405237913131714,
      "learning_rate": 7.016365973170208e-07,
      "loss": 0.4532,
      "step": 5555
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3445289134979248,
      "learning_rate": 7.003943713214422e-07,
      "loss": 0.4407,
      "step": 5556
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.431352138519287,
      "learning_rate": 6.991531631360338e-07,
      "loss": 0.4859,
      "step": 5557
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5351885557174683,
      "learning_rate": 6.979129730546191e-07,
      "loss": 0.4328,
      "step": 5558
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2381113767623901,
      "learning_rate": 6.966738013707774e-07,
      "loss": 0.4671,
      "step": 5559
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1632931232452393,
      "learning_rate": 6.954356483778457e-07,
      "loss": 0.4884,
      "step": 5560
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4585072994232178,
      "learning_rate": 6.941985143689239e-07,
      "loss": 0.4733,
      "step": 5561
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3478718996047974,
      "learning_rate": 6.929623996368673e-07,
      "loss": 0.4608,
      "step": 5562
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.289734959602356,
      "learning_rate": 6.917273044742928e-07,
      "loss": 0.5294,
      "step": 5563
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2730916738510132,
      "learning_rate": 6.904932291735739e-07,
      "loss": 0.3827,
      "step": 5564
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4382909536361694,
      "learning_rate": 6.892601740268406e-07,
      "loss": 0.518,
      "step": 5565
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7299211025238037,
      "learning_rate": 6.880281393259869e-07,
      "loss": 0.4257,
      "step": 5566
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3987921476364136,
      "learning_rate": 6.867971253626599e-07,
      "loss": 0.4813,
      "step": 5567
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9981815814971924,
      "learning_rate": 6.855671324282697e-07,
      "loss": 0.4569,
      "step": 5568
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4126816987991333,
      "learning_rate": 6.843381608139798e-07,
      "loss": 0.4904,
      "step": 5569
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2219067811965942,
      "learning_rate": 6.831102108107174e-07,
      "loss": 0.4512,
      "step": 5570
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5229506492614746,
      "learning_rate": 6.818832827091632e-07,
      "loss": 0.4787,
      "step": 5571
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2607550621032715,
      "learning_rate": 6.806573767997555e-07,
      "loss": 0.6772,
      "step": 5572
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1384875774383545,
      "learning_rate": 6.794324933726976e-07,
      "loss": 0.4981,
      "step": 5573
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3605701923370361,
      "learning_rate": 6.782086327179432e-07,
      "loss": 0.4592,
      "step": 5574
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.479987621307373,
      "learning_rate": 6.769857951252078e-07,
      "loss": 0.4835,
      "step": 5575
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4670110940933228,
      "learning_rate": 6.757639808839638e-07,
      "loss": 0.5124,
      "step": 5576
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3740513324737549,
      "learning_rate": 6.745431902834393e-07,
      "loss": 0.4173,
      "step": 5577
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3022741079330444,
      "learning_rate": 6.733234236126245e-07,
      "loss": 0.4755,
      "step": 5578
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8456859588623047,
      "learning_rate": 6.721046811602622e-07,
      "loss": 0.4455,
      "step": 5579
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7269436120986938,
      "learning_rate": 6.708869632148579e-07,
      "loss": 0.46,
      "step": 5580
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2340222597122192,
      "learning_rate": 6.696702700646695e-07,
      "loss": 0.4864,
      "step": 5581
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0697829723358154,
      "learning_rate": 6.684546019977167e-07,
      "loss": 0.4864,
      "step": 5582
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.298478364944458,
      "learning_rate": 6.672399593017726e-07,
      "loss": 0.4615,
      "step": 5583
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0728249549865723,
      "learning_rate": 6.66026342264371e-07,
      "loss": 0.4235,
      "step": 5584
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7965891361236572,
      "learning_rate": 6.64813751172802e-07,
      "loss": 0.4686,
      "step": 5585
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.438857913017273,
      "learning_rate": 6.636021863141101e-07,
      "loss": 0.4061,
      "step": 5586
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4805978536605835,
      "learning_rate": 6.623916479751013e-07,
      "loss": 0.512,
      "step": 5587
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.5901596546173096,
      "learning_rate": 6.61182136442336e-07,
      "loss": 0.4316,
      "step": 5588
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5982141494750977,
      "learning_rate": 6.599736520021299e-07,
      "loss": 0.4509,
      "step": 5589
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2472249269485474,
      "learning_rate": 6.587661949405599e-07,
      "loss": 0.3918,
      "step": 5590
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4759972095489502,
      "learning_rate": 6.575597655434551e-07,
      "loss": 0.3901,
      "step": 5591
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4059292078018188,
      "learning_rate": 6.563543640964059e-07,
      "loss": 0.47,
      "step": 5592
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.866397500038147,
      "learning_rate": 6.551499908847553e-07,
      "loss": 0.4652,
      "step": 5593
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2117769718170166,
      "learning_rate": 6.539466461936051e-07,
      "loss": 0.7047,
      "step": 5594
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3025590181350708,
      "learning_rate": 6.527443303078146e-07,
      "loss": 0.4867,
      "step": 5595
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2041957378387451,
      "learning_rate": 6.515430435119957e-07,
      "loss": 0.4811,
      "step": 5596
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5389330387115479,
      "learning_rate": 6.503427860905215e-07,
      "loss": 0.451,
      "step": 5597
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0356090068817139,
      "learning_rate": 6.491435583275168e-07,
      "loss": 0.4251,
      "step": 5598
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2735720872879028,
      "learning_rate": 6.479453605068676e-07,
      "loss": 0.4188,
      "step": 5599
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.442915916442871,
      "learning_rate": 6.467481929122111e-07,
      "loss": 0.4471,
      "step": 5600
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2537941932678223,
      "learning_rate": 6.455520558269423e-07,
      "loss": 0.4508,
      "step": 5601
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.399072289466858,
      "learning_rate": 6.443569495342156e-07,
      "loss": 0.4091,
      "step": 5602
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.706931710243225,
      "learning_rate": 6.431628743169354e-07,
      "loss": 0.4333,
      "step": 5603
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9967467188835144,
      "learning_rate": 6.419698304577687e-07,
      "loss": 0.3929,
      "step": 5604
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.305187702178955,
      "learning_rate": 6.407778182391322e-07,
      "loss": 0.472,
      "step": 5605
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4578882455825806,
      "learning_rate": 6.395868379432024e-07,
      "loss": 0.4178,
      "step": 5606
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4224779605865479,
      "learning_rate": 6.383968898519111e-07,
      "loss": 0.4447,
      "step": 5607
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5679636001586914,
      "learning_rate": 6.372079742469422e-07,
      "loss": 0.4488,
      "step": 5608
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4035325050354004,
      "learning_rate": 6.360200914097408e-07,
      "loss": 0.442,
      "step": 5609
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4940598011016846,
      "learning_rate": 6.348332416215025e-07,
      "loss": 0.4172,
      "step": 5610
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0817919969558716,
      "learning_rate": 6.336474251631824e-07,
      "loss": 0.4703,
      "step": 5611
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.318199634552002,
      "learning_rate": 6.324626423154867e-07,
      "loss": 0.4564,
      "step": 5612
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1971505880355835,
      "learning_rate": 6.312788933588815e-07,
      "loss": 0.4798,
      "step": 5613
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.159420132637024,
      "learning_rate": 6.300961785735849e-07,
      "loss": 0.4942,
      "step": 5614
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3759297132492065,
      "learning_rate": 6.2891449823957e-07,
      "loss": 0.4252,
      "step": 5615
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5609084367752075,
      "learning_rate": 6.277338526365689e-07,
      "loss": 0.4561,
      "step": 5616
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.3462018966674805,
      "learning_rate": 6.265542420440635e-07,
      "loss": 0.514,
      "step": 5617
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4641526937484741,
      "learning_rate": 6.253756667412947e-07,
      "loss": 0.4279,
      "step": 5618
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2618906497955322,
      "learning_rate": 6.241981270072572e-07,
      "loss": 0.514,
      "step": 5619
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2484989166259766,
      "learning_rate": 6.230216231206992e-07,
      "loss": 0.4319,
      "step": 5620
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4183225631713867,
      "learning_rate": 6.218461553601263e-07,
      "loss": 0.5193,
      "step": 5621
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.33867609500885,
      "learning_rate": 6.206717240037946e-07,
      "loss": 0.4731,
      "step": 5622
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.303782343864441,
      "learning_rate": 6.194983293297207e-07,
      "loss": 0.473,
      "step": 5623
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.438197374343872,
      "learning_rate": 6.183259716156703e-07,
      "loss": 0.4143,
      "step": 5624
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1250643730163574,
      "learning_rate": 6.17154651139168e-07,
      "loss": 0.4668,
      "step": 5625
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.356245517730713,
      "learning_rate": 6.15984368177489e-07,
      "loss": 0.4495,
      "step": 5626
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3230494260787964,
      "learning_rate": 6.148151230076648e-07,
      "loss": 0.5245,
      "step": 5627
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.47200345993042,
      "learning_rate": 6.136469159064817e-07,
      "loss": 0.4453,
      "step": 5628
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.540408730506897,
      "learning_rate": 6.124797471504795e-07,
      "loss": 0.4966,
      "step": 5629
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4800525903701782,
      "learning_rate": 6.113136170159545e-07,
      "loss": 0.4561,
      "step": 5630
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3927968740463257,
      "learning_rate": 6.10148525778953e-07,
      "loss": 0.4838,
      "step": 5631
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.426339030265808,
      "learning_rate": 6.089844737152761e-07,
      "loss": 0.4823,
      "step": 5632
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3750978708267212,
      "learning_rate": 6.078214611004835e-07,
      "loss": 0.4261,
      "step": 5633
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3170149326324463,
      "learning_rate": 6.066594882098831e-07,
      "loss": 0.4832,
      "step": 5634
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5576531887054443,
      "learning_rate": 6.054985553185405e-07,
      "loss": 0.4525,
      "step": 5635
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7508885860443115,
      "learning_rate": 6.043386627012726e-07,
      "loss": 0.3825,
      "step": 5636
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2648365497589111,
      "learning_rate": 6.031798106326525e-07,
      "loss": 0.4947,
      "step": 5637
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2885245084762573,
      "learning_rate": 6.020219993870057e-07,
      "loss": 0.4628,
      "step": 5638
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.611286163330078,
      "learning_rate": 6.008652292384081e-07,
      "loss": 0.4305,
      "step": 5639
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.580264687538147,
      "learning_rate": 5.997095004606973e-07,
      "loss": 0.4248,
      "step": 5640
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3984488248825073,
      "learning_rate": 5.98554813327456e-07,
      "loss": 0.4845,
      "step": 5641
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.515753149986267,
      "learning_rate": 5.974011681120256e-07,
      "loss": 0.4511,
      "step": 5642
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4304625988006592,
      "learning_rate": 5.962485650874988e-07,
      "loss": 0.4274,
      "step": 5643
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2829954624176025,
      "learning_rate": 5.950970045267196e-07,
      "loss": 0.5118,
      "step": 5644
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6260404586791992,
      "learning_rate": 5.939464867022909e-07,
      "loss": 0.4076,
      "step": 5645
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3375844955444336,
      "learning_rate": 5.927970118865617e-07,
      "loss": 0.5181,
      "step": 5646
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1287881135940552,
      "learning_rate": 5.916485803516409e-07,
      "loss": 0.4675,
      "step": 5647
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.219273567199707,
      "learning_rate": 5.905011923693848e-07,
      "loss": 0.4321,
      "step": 5648
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3860880136489868,
      "learning_rate": 5.89354848211407e-07,
      "loss": 0.4405,
      "step": 5649
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6731235980987549,
      "learning_rate": 5.882095481490701e-07,
      "loss": 0.4632,
      "step": 5650
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2018948793411255,
      "learning_rate": 5.870652924534925e-07,
      "loss": 0.48,
      "step": 5651
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1154621839523315,
      "learning_rate": 5.859220813955458e-07,
      "loss": 0.4465,
      "step": 5652
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.252830982208252,
      "learning_rate": 5.847799152458506e-07,
      "loss": 0.4882,
      "step": 5653
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.223337173461914,
      "learning_rate": 5.836387942747845e-07,
      "loss": 0.4805,
      "step": 5654
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4134505987167358,
      "learning_rate": 5.82498718752475e-07,
      "loss": 0.4519,
      "step": 5655
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2190040349960327,
      "learning_rate": 5.81359688948801e-07,
      "loss": 0.4161,
      "step": 5656
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9428551197052002,
      "learning_rate": 5.802217051333981e-07,
      "loss": 0.4721,
      "step": 5657
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1818989515304565,
      "learning_rate": 5.7908476757565e-07,
      "loss": 0.7162,
      "step": 5658
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1742883920669556,
      "learning_rate": 5.779488765446967e-07,
      "loss": 0.3943,
      "step": 5659
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.1071207523345947,
      "learning_rate": 5.768140323094252e-07,
      "loss": 0.4193,
      "step": 5660
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3579511642456055,
      "learning_rate": 5.7568023513848e-07,
      "loss": 0.4964,
      "step": 5661
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1985818147659302,
      "learning_rate": 5.745474853002564e-07,
      "loss": 0.462,
      "step": 5662
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2558389902114868,
      "learning_rate": 5.734157830628984e-07,
      "loss": 0.4668,
      "step": 5663
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.894976615905762,
      "learning_rate": 5.722851286943065e-07,
      "loss": 0.4819,
      "step": 5664
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1930205821990967,
      "learning_rate": 5.711555224621296e-07,
      "loss": 0.4077,
      "step": 5665
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6000336408615112,
      "learning_rate": 5.700269646337725e-07,
      "loss": 0.476,
      "step": 5666
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3615154027938843,
      "learning_rate": 5.688994554763871e-07,
      "loss": 0.4691,
      "step": 5667
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4035141468048096,
      "learning_rate": 5.677729952568794e-07,
      "loss": 0.4336,
      "step": 5668
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0767536163330078,
      "learning_rate": 5.666475842419089e-07,
      "loss": 0.6799,
      "step": 5669
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4349490404129028,
      "learning_rate": 5.655232226978824e-07,
      "loss": 0.5001,
      "step": 5670
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4895105361938477,
      "learning_rate": 5.643999108909631e-07,
      "loss": 0.4548,
      "step": 5671
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1921666860580444,
      "learning_rate": 5.632776490870612e-07,
      "loss": 0.4603,
      "step": 5672
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.271531343460083,
      "learning_rate": 5.621564375518413e-07,
      "loss": 0.4499,
      "step": 5673
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.380265474319458,
      "learning_rate": 5.6103627655072e-07,
      "loss": 0.4755,
      "step": 5674
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9582182168960571,
      "learning_rate": 5.599171663488617e-07,
      "loss": 0.4363,
      "step": 5675
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6339881420135498,
      "learning_rate": 5.58799107211186e-07,
      "loss": 0.4737,
      "step": 5676
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3990132808685303,
      "learning_rate": 5.576820994023596e-07,
      "loss": 0.4198,
      "step": 5677
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1462196111679077,
      "learning_rate": 5.565661431868046e-07,
      "loss": 0.3771,
      "step": 5678
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0597957372665405,
      "learning_rate": 5.554512388286915e-07,
      "loss": 0.6896,
      "step": 5679
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.100401520729065,
      "learning_rate": 5.543373865919411e-07,
      "loss": 0.4472,
      "step": 5680
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.500095009803772,
      "learning_rate": 5.53224586740228e-07,
      "loss": 0.4627,
      "step": 5681
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.253967046737671,
      "learning_rate": 5.521128395369751e-07,
      "loss": 0.5086,
      "step": 5682
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.095141887664795,
      "learning_rate": 5.510021452453579e-07,
      "loss": 0.4151,
      "step": 5683
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3069007396697998,
      "learning_rate": 5.498925041283027e-07,
      "loss": 0.4687,
      "step": 5684
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.470335602760315,
      "learning_rate": 5.487839164484831e-07,
      "loss": 0.4043,
      "step": 5685
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.155324101448059,
      "learning_rate": 5.47676382468329e-07,
      "loss": 0.4419,
      "step": 5686
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.7277793884277344,
      "learning_rate": 5.465699024500148e-07,
      "loss": 0.5023,
      "step": 5687
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5163142681121826,
      "learning_rate": 5.454644766554712e-07,
      "loss": 0.4547,
      "step": 5688
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2228119373321533,
      "learning_rate": 5.443601053463743e-07,
      "loss": 0.42,
      "step": 5689
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.6846420764923096,
      "learning_rate": 5.432567887841544e-07,
      "loss": 0.4232,
      "step": 5690
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2179888486862183,
      "learning_rate": 5.4215452722999e-07,
      "loss": 0.4241,
      "step": 5691
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0593209266662598,
      "learning_rate": 5.410533209448088e-07,
      "loss": 0.4421,
      "step": 5692
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0566012859344482,
      "learning_rate": 5.39953170189293e-07,
      "loss": 0.4547,
      "step": 5693
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.502929449081421,
      "learning_rate": 5.388540752238703e-07,
      "loss": 0.4873,
      "step": 5694
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2886162996292114,
      "learning_rate": 5.377560363087204e-07,
      "loss": 0.4697,
      "step": 5695
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9038890600204468,
      "learning_rate": 5.366590537037747e-07,
      "loss": 0.4599,
      "step": 5696
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5626318454742432,
      "learning_rate": 5.355631276687107e-07,
      "loss": 0.4684,
      "step": 5697
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7095704078674316,
      "learning_rate": 5.344682584629601e-07,
      "loss": 0.4569,
      "step": 5698
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9013302326202393,
      "learning_rate": 5.333744463456997e-07,
      "loss": 0.4889,
      "step": 5699
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2365692853927612,
      "learning_rate": 5.322816915758616e-07,
      "loss": 0.4494,
      "step": 5700
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5409573316574097,
      "learning_rate": 5.311899944121218e-07,
      "loss": 0.3824,
      "step": 5701
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.245675802230835,
      "learning_rate": 5.300993551129108e-07,
      "loss": 0.4235,
      "step": 5702
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3354871273040771,
      "learning_rate": 5.290097739364064e-07,
      "loss": 0.4562,
      "step": 5703
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5839320421218872,
      "learning_rate": 5.279212511405341e-07,
      "loss": 0.3966,
      "step": 5704
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0347988605499268,
      "learning_rate": 5.268337869829737e-07,
      "loss": 0.6675,
      "step": 5705
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.232500433921814,
      "learning_rate": 5.257473817211483e-07,
      "loss": 0.4434,
      "step": 5706
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2043354511260986,
      "learning_rate": 5.246620356122378e-07,
      "loss": 0.4936,
      "step": 5707
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2181463241577148,
      "learning_rate": 5.23577748913165e-07,
      "loss": 0.4228,
      "step": 5708
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.324017882347107,
      "learning_rate": 5.224945218806032e-07,
      "loss": 0.477,
      "step": 5709
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2171804904937744,
      "learning_rate": 5.214123547709777e-07,
      "loss": 0.5018,
      "step": 5710
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0757993459701538,
      "learning_rate": 5.203312478404598e-07,
      "loss": 0.4695,
      "step": 5711
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2929309606552124,
      "learning_rate": 5.192512013449719e-07,
      "loss": 0.3843,
      "step": 5712
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.695616602897644,
      "learning_rate": 5.181722155401831e-07,
      "loss": 0.4877,
      "step": 5713
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1144698858261108,
      "learning_rate": 5.170942906815152e-07,
      "loss": 0.4134,
      "step": 5714
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2532333135604858,
      "learning_rate": 5.160174270241341e-07,
      "loss": 0.4301,
      "step": 5715
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3023666143417358,
      "learning_rate": 5.149416248229578e-07,
      "loss": 0.5103,
      "step": 5716
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0621356964111328,
      "learning_rate": 5.138668843326511e-07,
      "loss": 0.5365,
      "step": 5717
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1373388767242432,
      "learning_rate": 5.127932058076296e-07,
      "loss": 0.4759,
      "step": 5718
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3188871145248413,
      "learning_rate": 5.117205895020571e-07,
      "loss": 0.4507,
      "step": 5719
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5364210605621338,
      "learning_rate": 5.106490356698446e-07,
      "loss": 0.4853,
      "step": 5720
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.162329077720642,
      "learning_rate": 5.095785445646506e-07,
      "loss": 0.4529,
      "step": 5721
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7131632566452026,
      "learning_rate": 5.085091164398853e-07,
      "loss": 0.4226,
      "step": 5722
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.396528720855713,
      "learning_rate": 5.074407515487051e-07,
      "loss": 0.4908,
      "step": 5723
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2368204593658447,
      "learning_rate": 5.063734501440154e-07,
      "loss": 0.4368,
      "step": 5724
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.5100486278533936,
      "learning_rate": 5.053072124784692e-07,
      "loss": 0.4872,
      "step": 5725
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.235329508781433,
      "learning_rate": 5.042420388044689e-07,
      "loss": 0.6809,
      "step": 5726
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2388979196548462,
      "learning_rate": 5.031779293741635e-07,
      "loss": 0.4526,
      "step": 5727
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4227555990219116,
      "learning_rate": 5.021148844394502e-07,
      "loss": 0.4568,
      "step": 5728
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.1915268898010254,
      "learning_rate": 5.010529042519779e-07,
      "loss": 0.5843,
      "step": 5729
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.197419285774231,
      "learning_rate": 4.999919890631366e-07,
      "loss": 0.4542,
      "step": 5730
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1312289237976074,
      "learning_rate": 4.989321391240709e-07,
      "loss": 0.4587,
      "step": 5731
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.1839160919189453,
      "learning_rate": 4.978733546856695e-07,
      "loss": 0.5065,
      "step": 5732
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4627065658569336,
      "learning_rate": 4.968156359985682e-07,
      "loss": 0.5505,
      "step": 5733
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5818848609924316,
      "learning_rate": 4.957589833131543e-07,
      "loss": 0.5327,
      "step": 5734
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2891411781311035,
      "learning_rate": 4.947033968795584e-07,
      "loss": 0.5514,
      "step": 5735
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6110657453536987,
      "learning_rate": 4.936488769476628e-07,
      "loss": 0.4776,
      "step": 5736
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5672775506973267,
      "learning_rate": 4.925954237670927e-07,
      "loss": 0.4934,
      "step": 5737
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4623671770095825,
      "learning_rate": 4.915430375872266e-07,
      "loss": 0.4443,
      "step": 5738
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4233046770095825,
      "learning_rate": 4.904917186571839e-07,
      "loss": 0.4334,
      "step": 5739
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1913989782333374,
      "learning_rate": 4.894414672258369e-07,
      "loss": 0.4,
      "step": 5740
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0776864290237427,
      "learning_rate": 4.883922835418032e-07,
      "loss": 0.4661,
      "step": 5741
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.335527181625366,
      "learning_rate": 4.873441678534457e-07,
      "loss": 0.489,
      "step": 5742
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1217743158340454,
      "learning_rate": 4.862971204088774e-07,
      "loss": 0.3755,
      "step": 5743
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4234479665756226,
      "learning_rate": 4.852511414559575e-07,
      "loss": 0.4239,
      "step": 5744
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9107964038848877,
      "learning_rate": 4.842062312422902e-07,
      "loss": 0.4204,
      "step": 5745
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.2015767097473145,
      "learning_rate": 4.831623900152304e-07,
      "loss": 0.3589,
      "step": 5746
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7560564279556274,
      "learning_rate": 4.821196180218762e-07,
      "loss": 0.4671,
      "step": 5747
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.7441024780273438,
      "learning_rate": 4.810779155090761e-07,
      "loss": 0.5306,
      "step": 5748
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5188394784927368,
      "learning_rate": 4.800372827234229e-07,
      "loss": 0.5067,
      "step": 5749
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.3031325340270996,
      "learning_rate": 4.789977199112567e-07,
      "loss": 0.4964,
      "step": 5750
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1710633039474487,
      "learning_rate": 4.779592273186656e-07,
      "loss": 0.681,
      "step": 5751
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3201124668121338,
      "learning_rate": 4.769218051914825e-07,
      "loss": 0.5031,
      "step": 5752
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.210752248764038,
      "learning_rate": 4.758854537752894e-07,
      "loss": 0.3837,
      "step": 5753
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2939445972442627,
      "learning_rate": 4.748501733154104e-07,
      "loss": 0.4511,
      "step": 5754
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3622833490371704,
      "learning_rate": 4.7381596405692174e-07,
      "loss": 0.3782,
      "step": 5755
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3254129886627197,
      "learning_rate": 4.72782826244641e-07,
      "loss": 0.47,
      "step": 5756
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.5483908653259277,
      "learning_rate": 4.717507601231364e-07,
      "loss": 0.4518,
      "step": 5757
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2435483932495117,
      "learning_rate": 4.707197659367191e-07,
      "loss": 0.3782,
      "step": 5758
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.541017770767212,
      "learning_rate": 4.6968984392944786e-07,
      "loss": 0.5076,
      "step": 5759
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.075971245765686,
      "learning_rate": 4.6866099434512815e-07,
      "loss": 0.4125,
      "step": 5760
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2125295400619507,
      "learning_rate": 4.6763321742731016e-07,
      "loss": 0.4592,
      "step": 5761
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.9819726943969727,
      "learning_rate": 4.6660651341929155e-07,
      "loss": 0.4953,
      "step": 5762
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.6482889652252197,
      "learning_rate": 4.6558088256411673e-07,
      "loss": 0.4067,
      "step": 5763
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4506549835205078,
      "learning_rate": 4.6455632510457313e-07,
      "loss": 0.5077,
      "step": 5764
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5879064798355103,
      "learning_rate": 4.6353284128319684e-07,
      "loss": 0.5106,
      "step": 5765
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.308320164680481,
      "learning_rate": 4.625104313422674e-07,
      "loss": 0.509,
      "step": 5766
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3769840002059937,
      "learning_rate": 4.6148909552381326e-07,
      "loss": 0.5145,
      "step": 5767
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0763161182403564,
      "learning_rate": 4.604688340696056e-07,
      "loss": 0.6641,
      "step": 5768
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6774725914001465,
      "learning_rate": 4.5944964722116294e-07,
      "loss": 0.4616,
      "step": 5769
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2145435810089111,
      "learning_rate": 4.584315352197494e-07,
      "loss": 0.4233,
      "step": 5770
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3187719583511353,
      "learning_rate": 4.5741449830637275e-07,
      "loss": 0.4672,
      "step": 5771
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.218598484992981,
      "learning_rate": 4.563985367217888e-07,
      "loss": 0.3414,
      "step": 5772
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2271095514297485,
      "learning_rate": 4.5538365070649795e-07,
      "loss": 0.4764,
      "step": 5773
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1854524612426758,
      "learning_rate": 4.543698405007452e-07,
      "loss": 0.4818,
      "step": 5774
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.422088623046875,
      "learning_rate": 4.5335710634452246e-07,
      "loss": 0.3779,
      "step": 5775
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2697899341583252,
      "learning_rate": 4.523454484775647e-07,
      "loss": 0.5128,
      "step": 5776
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.185440182685852,
      "learning_rate": 4.5133486713935416e-07,
      "loss": 0.4671,
      "step": 5777
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1013423204421997,
      "learning_rate": 4.503253625691167e-07,
      "loss": 0.4551,
      "step": 5778
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4917235374450684,
      "learning_rate": 4.493169350058252e-07,
      "loss": 0.4334,
      "step": 5779
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2373337745666504,
      "learning_rate": 4.483095846881946e-07,
      "loss": 0.4887,
      "step": 5780
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3173052072525024,
      "learning_rate": 4.473033118546893e-07,
      "loss": 0.4888,
      "step": 5781
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.396897554397583,
      "learning_rate": 4.4629811674351373e-07,
      "loss": 0.4972,
      "step": 5782
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.306300163269043,
      "learning_rate": 4.4529399959261923e-07,
      "loss": 0.422,
      "step": 5783
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4503238201141357,
      "learning_rate": 4.442909606397033e-07,
      "loss": 0.4621,
      "step": 5784
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2649452686309814,
      "learning_rate": 4.4328900012220623e-07,
      "loss": 0.4819,
      "step": 5785
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.142683506011963,
      "learning_rate": 4.4228811827731586e-07,
      "loss": 0.5148,
      "step": 5786
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.199125051498413,
      "learning_rate": 4.4128831534196116e-07,
      "loss": 0.4336,
      "step": 5787
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4011729955673218,
      "learning_rate": 4.4028959155281603e-07,
      "loss": 0.4812,
      "step": 5788
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3489487171173096,
      "learning_rate": 4.3929194714630306e-07,
      "loss": 0.4741,
      "step": 5789
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.87539541721344,
      "learning_rate": 4.382953823585834e-07,
      "loss": 0.4874,
      "step": 5790
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6443650722503662,
      "learning_rate": 4.3729989742556765e-07,
      "loss": 0.4834,
      "step": 5791
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.232426643371582,
      "learning_rate": 4.3630549258290745e-07,
      "loss": 0.4544,
      "step": 5792
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4299677610397339,
      "learning_rate": 4.3531216806600105e-07,
      "loss": 0.4545,
      "step": 5793
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3271969556808472,
      "learning_rate": 4.343199241099905e-07,
      "loss": 0.4936,
      "step": 5794
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2022013664245605,
      "learning_rate": 4.3332876094975775e-07,
      "loss": 0.4599,
      "step": 5795
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4878846406936646,
      "learning_rate": 4.3233867881993794e-07,
      "loss": 0.4039,
      "step": 5796
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3133450746536255,
      "learning_rate": 4.313496779549015e-07,
      "loss": 0.4422,
      "step": 5797
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1409324407577515,
      "learning_rate": 4.303617585887687e-07,
      "loss": 0.4674,
      "step": 5798
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.077069878578186,
      "learning_rate": 4.2937492095540043e-07,
      "loss": 0.4524,
      "step": 5799
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.259340524673462,
      "learning_rate": 4.2838916528840146e-07,
      "loss": 0.4596,
      "step": 5800
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3158483505249023,
      "learning_rate": 4.2740449182112344e-07,
      "loss": 0.4714,
      "step": 5801
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5187327861785889,
      "learning_rate": 4.264209007866588e-07,
      "loss": 0.5812,
      "step": 5802
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.5151965618133545,
      "learning_rate": 4.2543839241784624e-07,
      "loss": 0.476,
      "step": 5803
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.193746328353882,
      "learning_rate": 4.2445696694726536e-07,
      "loss": 0.4576,
      "step": 5804
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7926398515701294,
      "learning_rate": 4.234766246072419e-07,
      "loss": 0.4518,
      "step": 5805
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.014349460601807,
      "learning_rate": 4.224973656298437e-07,
      "loss": 0.4887,
      "step": 5806
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6048614978790283,
      "learning_rate": 4.2151919024688237e-07,
      "loss": 0.4301,
      "step": 5807
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4350149631500244,
      "learning_rate": 4.20542098689915e-07,
      "loss": 0.4437,
      "step": 5808
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.033567190170288,
      "learning_rate": 4.1956609119023825e-07,
      "loss": 0.3939,
      "step": 5809
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6192089319229126,
      "learning_rate": 4.185911679788962e-07,
      "loss": 0.4258,
      "step": 5810
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4109331369400024,
      "learning_rate": 4.176173292866731e-07,
      "loss": 0.4996,
      "step": 5811
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3264445066452026,
      "learning_rate": 4.1664457534409785e-07,
      "loss": 0.4494,
      "step": 5812
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1500216722488403,
      "learning_rate": 4.156729063814435e-07,
      "loss": 0.6906,
      "step": 5813
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.719008207321167,
      "learning_rate": 4.1470232262872337e-07,
      "loss": 0.4345,
      "step": 5814
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1430816650390625,
      "learning_rate": 4.1373282431569793e-07,
      "loss": 0.4168,
      "step": 5815
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4510760307312012,
      "learning_rate": 4.1276441167186596e-07,
      "loss": 0.4426,
      "step": 5816
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9908740520477295,
      "learning_rate": 4.117970849264741e-07,
      "loss": 0.452,
      "step": 5817
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1533371210098267,
      "learning_rate": 4.1083084430850906e-07,
      "loss": 0.4772,
      "step": 5818
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2875419855117798,
      "learning_rate": 4.098656900467002e-07,
      "loss": 0.407,
      "step": 5819
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1631487607955933,
      "learning_rate": 4.089016223695219e-07,
      "loss": 0.4528,
      "step": 5820
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0777612924575806,
      "learning_rate": 4.079386415051884e-07,
      "loss": 0.4077,
      "step": 5821
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2667567729949951,
      "learning_rate": 4.069767476816594e-07,
      "loss": 0.4309,
      "step": 5822
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.2752747535705566,
      "learning_rate": 4.060159411266362e-07,
      "loss": 0.4023,
      "step": 5823
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5777182579040527,
      "learning_rate": 4.050562220675608e-07,
      "loss": 0.4152,
      "step": 5824
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.158079743385315,
      "learning_rate": 4.040975907316219e-07,
      "loss": 0.4576,
      "step": 5825
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.242693543434143,
      "learning_rate": 4.031400473457464e-07,
      "loss": 0.3705,
      "step": 5826
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1734428405761719,
      "learning_rate": 4.0218359213660685e-07,
      "loss": 0.4264,
      "step": 5827
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.140810012817383,
      "learning_rate": 4.0122822533061657e-07,
      "loss": 0.4491,
      "step": 5828
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1072782278060913,
      "learning_rate": 4.002739471539313e-07,
      "loss": 0.4196,
      "step": 5829
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7242637872695923,
      "learning_rate": 3.993207578324515e-07,
      "loss": 0.4827,
      "step": 5830
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2582381963729858,
      "learning_rate": 3.9836865759181444e-07,
      "loss": 0.3543,
      "step": 5831
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.282475471496582,
      "learning_rate": 3.9741764665740647e-07,
      "loss": 0.4541,
      "step": 5832
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2435204982757568,
      "learning_rate": 3.964677252543497e-07,
      "loss": 0.5227,
      "step": 5833
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.207763671875,
      "learning_rate": 3.9551889360751325e-07,
      "loss": 0.4073,
      "step": 5834
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.455489993095398,
      "learning_rate": 3.945711519415052e-07,
      "loss": 0.4879,
      "step": 5835
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4915642738342285,
      "learning_rate": 3.9362450048067544e-07,
      "loss": 0.5297,
      "step": 5836
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.117576241493225,
      "learning_rate": 3.9267893944911985e-07,
      "loss": 0.6572,
      "step": 5837
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2709693908691406,
      "learning_rate": 3.917344690706704e-07,
      "loss": 0.4561,
      "step": 5838
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.238675832748413,
      "learning_rate": 3.9079108956890544e-07,
      "loss": 0.4266,
      "step": 5839
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3851293325424194,
      "learning_rate": 3.8984880116714364e-07,
      "loss": 0.4424,
      "step": 5840
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3219335079193115,
      "learning_rate": 3.8890760408844374e-07,
      "loss": 0.5007,
      "step": 5841
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3074020147323608,
      "learning_rate": 3.879674985556092e-07,
      "loss": 0.4509,
      "step": 5842
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3441299200057983,
      "learning_rate": 3.870284847911815e-07,
      "loss": 0.4752,
      "step": 5843
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5932401418685913,
      "learning_rate": 3.8609056301744773e-07,
      "loss": 0.4367,
      "step": 5844
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0920616388320923,
      "learning_rate": 3.8515373345643315e-07,
      "loss": 0.4137,
      "step": 5845
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.534996747970581,
      "learning_rate": 3.8421799632990585e-07,
      "loss": 0.4144,
      "step": 5846
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6597113609313965,
      "learning_rate": 3.832833518593759e-07,
      "loss": 0.5243,
      "step": 5847
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4554145336151123,
      "learning_rate": 3.823498002660925e-07,
      "loss": 0.4156,
      "step": 5848
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.224661946296692,
      "learning_rate": 3.814173417710487e-07,
      "loss": 0.5112,
      "step": 5849
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3760015964508057,
      "learning_rate": 3.8048597659497643e-07,
      "loss": 0.4689,
      "step": 5850
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3549566268920898,
      "learning_rate": 3.795557049583526e-07,
      "loss": 0.4662,
      "step": 5851
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3608874082565308,
      "learning_rate": 3.78626527081391e-07,
      "loss": 0.4532,
      "step": 5852
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7491191625595093,
      "learning_rate": 3.7769844318404805e-07,
      "loss": 0.4466,
      "step": 5853
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3939679861068726,
      "learning_rate": 3.767714534860223e-07,
      "loss": 0.5106,
      "step": 5854
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2114839553833008,
      "learning_rate": 3.7584555820675106e-07,
      "loss": 0.499,
      "step": 5855
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7198859453201294,
      "learning_rate": 3.749207575654157e-07,
      "loss": 0.4733,
      "step": 5856
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9537566900253296,
      "learning_rate": 3.7399705178093493e-07,
      "loss": 0.4527,
      "step": 5857
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3337112665176392,
      "learning_rate": 3.7307444107197113e-07,
      "loss": 0.4732,
      "step": 5858
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2732981443405151,
      "learning_rate": 3.7215292565692617e-07,
      "loss": 0.5136,
      "step": 5859
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7222964763641357,
      "learning_rate": 3.7123250575394234e-07,
      "loss": 0.4447,
      "step": 5860
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.493822455406189,
      "learning_rate": 3.7031318158090354e-07,
      "loss": 0.4355,
      "step": 5861
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4089196920394897,
      "learning_rate": 3.693949533554314e-07,
      "loss": 0.4635,
      "step": 5862
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9020841121673584,
      "learning_rate": 3.6847782129489527e-07,
      "loss": 0.4925,
      "step": 5863
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2926968336105347,
      "learning_rate": 3.675617856163971e-07,
      "loss": 0.4376,
      "step": 5864
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.430238127708435,
      "learning_rate": 3.6664684653678226e-07,
      "loss": 0.4484,
      "step": 5865
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1187047958374023,
      "learning_rate": 3.657330042726381e-07,
      "loss": 0.4451,
      "step": 5866
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3368456363677979,
      "learning_rate": 3.648202590402894e-07,
      "loss": 0.4671,
      "step": 5867
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1524295806884766,
      "learning_rate": 3.639086110558049e-07,
      "loss": 0.438,
      "step": 5868
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4600518941879272,
      "learning_rate": 3.629980605349898e-07,
      "loss": 0.4637,
      "step": 5869
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1417014598846436,
      "learning_rate": 3.6208860769339214e-07,
      "loss": 0.4081,
      "step": 5870
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0413745641708374,
      "learning_rate": 3.611802527462976e-07,
      "loss": 0.4109,
      "step": 5871
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.424892783164978,
      "learning_rate": 3.6027299590873643e-07,
      "loss": 0.3961,
      "step": 5872
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0995979309082031,
      "learning_rate": 3.5936683739547284e-07,
      "loss": 0.4644,
      "step": 5873
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4711461067199707,
      "learning_rate": 3.5846177742101586e-07,
      "loss": 0.4863,
      "step": 5874
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9905712604522705,
      "learning_rate": 3.5755781619961406e-07,
      "loss": 0.3897,
      "step": 5875
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1393564939498901,
      "learning_rate": 3.5665495394525296e-07,
      "loss": 0.5068,
      "step": 5876
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2730227708816528,
      "learning_rate": 3.557531908716599e-07,
      "loss": 0.4697,
      "step": 5877
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2355456352233887,
      "learning_rate": 3.5485252719230256e-07,
      "loss": 0.4493,
      "step": 5878
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7356599569320679,
      "learning_rate": 3.5395296312038643e-07,
      "loss": 0.4072,
      "step": 5879
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.4974257946014404,
      "learning_rate": 3.5305449886886e-07,
      "loss": 0.4396,
      "step": 5880
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.666938066482544,
      "learning_rate": 3.5215713465040656e-07,
      "loss": 0.5726,
      "step": 5881
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3037830591201782,
      "learning_rate": 3.5126087067745395e-07,
      "loss": 0.4583,
      "step": 5882
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3636102676391602,
      "learning_rate": 3.503657071621658e-07,
      "loss": 0.4588,
      "step": 5883
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.707503080368042,
      "learning_rate": 3.4947164431644754e-07,
      "loss": 0.5064,
      "step": 5884
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0563658475875854,
      "learning_rate": 3.4857868235194383e-07,
      "loss": 0.679,
      "step": 5885
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2954094409942627,
      "learning_rate": 3.476868214800372e-07,
      "loss": 0.4395,
      "step": 5886
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4975076913833618,
      "learning_rate": 3.467960619118516e-07,
      "loss": 0.4112,
      "step": 5887
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2981542348861694,
      "learning_rate": 3.459064038582482e-07,
      "loss": 0.4826,
      "step": 5888
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5553066730499268,
      "learning_rate": 3.4501784752982824e-07,
      "loss": 0.4939,
      "step": 5889
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0558820962905884,
      "learning_rate": 3.441303931369339e-07,
      "loss": 0.4349,
      "step": 5890
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4594128131866455,
      "learning_rate": 3.432440408896426e-07,
      "loss": 0.493,
      "step": 5891
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1243460178375244,
      "learning_rate": 3.423587909977755e-07,
      "loss": 0.4379,
      "step": 5892
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3282485008239746,
      "learning_rate": 3.414746436708882e-07,
      "loss": 0.4712,
      "step": 5893
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.197563886642456,
      "learning_rate": 3.405915991182801e-07,
      "loss": 0.4914,
      "step": 5894
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1407170295715332,
      "learning_rate": 3.39709657548985e-07,
      "loss": 0.6459,
      "step": 5895
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8801995515823364,
      "learning_rate": 3.388288191717787e-07,
      "loss": 0.5124,
      "step": 5896
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2862392663955688,
      "learning_rate": 3.379490841951749e-07,
      "loss": 0.4201,
      "step": 5897
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.1583940982818604,
      "learning_rate": 3.370704528274249e-07,
      "loss": 0.4234,
      "step": 5898
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2676560878753662,
      "learning_rate": 3.361929252765217e-07,
      "loss": 0.4694,
      "step": 5899
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1649049520492554,
      "learning_rate": 3.353165017501936e-07,
      "loss": 0.4934,
      "step": 5900
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2580654621124268,
      "learning_rate": 3.3444118245590963e-07,
      "loss": 0.4742,
      "step": 5901
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9891720414161682,
      "learning_rate": 3.335669676008774e-07,
      "loss": 0.7063,
      "step": 5902
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4388841390609741,
      "learning_rate": 3.326938573920413e-07,
      "loss": 0.5648,
      "step": 5903
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.629974842071533,
      "learning_rate": 3.3182185203608663e-07,
      "loss": 0.4847,
      "step": 5904
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.1329939365386963,
      "learning_rate": 3.3095095173943545e-07,
      "loss": 0.4749,
      "step": 5905
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2844605445861816,
      "learning_rate": 3.3008115670824846e-07,
      "loss": 0.47,
      "step": 5906
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1941887140274048,
      "learning_rate": 3.2921246714842704e-07,
      "loss": 0.421,
      "step": 5907
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6689413785934448,
      "learning_rate": 3.283448832656061e-07,
      "loss": 0.4616,
      "step": 5908
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0877468585968018,
      "learning_rate": 3.2747840526516414e-07,
      "loss": 0.4416,
      "step": 5909
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.260217308998108,
      "learning_rate": 3.2661303335221327e-07,
      "loss": 0.473,
      "step": 5910
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1298803091049194,
      "learning_rate": 3.2574876773160733e-07,
      "loss": 0.3995,
      "step": 5911
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2351245880126953,
      "learning_rate": 3.2488560860793603e-07,
      "loss": 0.468,
      "step": 5912
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6919783353805542,
      "learning_rate": 3.2402355618552816e-07,
      "loss": 0.4579,
      "step": 5913
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.311331868171692,
      "learning_rate": 3.23162610668451e-07,
      "loss": 0.4578,
      "step": 5914
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6386631727218628,
      "learning_rate": 3.223027722605071e-07,
      "loss": 0.4269,
      "step": 5915
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3772228956222534,
      "learning_rate": 3.214440411652409e-07,
      "loss": 0.5092,
      "step": 5916
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3654600381851196,
      "learning_rate": 3.2058641758593135e-07,
      "loss": 0.4609,
      "step": 5917
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7394691705703735,
      "learning_rate": 3.1972990172559725e-07,
      "loss": 0.4271,
      "step": 5918
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.3841307163238525,
      "learning_rate": 3.188744937869953e-07,
      "loss": 0.3872,
      "step": 5919
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2226077318191528,
      "learning_rate": 3.180201939726174e-07,
      "loss": 0.448,
      "step": 5920
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1775826215744019,
      "learning_rate": 3.1716700248469677e-07,
      "loss": 0.4259,
      "step": 5921
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4950835704803467,
      "learning_rate": 3.163149195252002e-07,
      "loss": 0.4476,
      "step": 5922
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.169851303100586,
      "learning_rate": 3.154639452958369e-07,
      "loss": 0.4895,
      "step": 5923
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2418406009674072,
      "learning_rate": 3.1461407999804903e-07,
      "loss": 0.4498,
      "step": 5924
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.206714153289795,
      "learning_rate": 3.1376532383301906e-07,
      "loss": 0.6886,
      "step": 5925
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5479193925857544,
      "learning_rate": 3.129176770016662e-07,
      "loss": 0.4875,
      "step": 5926
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0861128568649292,
      "learning_rate": 3.12071139704645e-07,
      "loss": 0.6958,
      "step": 5927
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2482603788375854,
      "learning_rate": 3.112257121423512e-07,
      "loss": 0.4247,
      "step": 5928
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.655537724494934,
      "learning_rate": 3.1038139451491533e-07,
      "loss": 0.5049,
      "step": 5929
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.9654793739318848,
      "learning_rate": 3.095381870222064e-07,
      "loss": 0.4676,
      "step": 5930
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3238675594329834,
      "learning_rate": 3.086960898638297e-07,
      "loss": 0.5129,
      "step": 5931
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2283321619033813,
      "learning_rate": 3.078551032391269e-07,
      "loss": 0.4765,
      "step": 5932
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6158150434494019,
      "learning_rate": 3.070152273471788e-07,
      "loss": 0.4112,
      "step": 5933
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2964568138122559,
      "learning_rate": 3.0617646238680187e-07,
      "loss": 0.4366,
      "step": 5934
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4795736074447632,
      "learning_rate": 3.0533880855655064e-07,
      "loss": 0.4701,
      "step": 5935
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.162533164024353,
      "learning_rate": 3.045022660547148e-07,
      "loss": 0.5017,
      "step": 5936
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5489883422851562,
      "learning_rate": 3.036668350793232e-07,
      "loss": 0.493,
      "step": 5937
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0848665237426758,
      "learning_rate": 3.0283251582814087e-07,
      "loss": 0.4387,
      "step": 5938
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3551691770553589,
      "learning_rate": 3.019993084986667e-07,
      "loss": 0.4568,
      "step": 5939
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.305694341659546,
      "learning_rate": 3.011672132881427e-07,
      "loss": 0.4105,
      "step": 5940
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1677476167678833,
      "learning_rate": 3.0033623039354144e-07,
      "loss": 0.4499,
      "step": 5941
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8759924173355103,
      "learning_rate": 2.9950636001157607e-07,
      "loss": 0.5182,
      "step": 5942
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1639631986618042,
      "learning_rate": 2.986776023386939e-07,
      "loss": 0.4611,
      "step": 5943
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.346375823020935,
      "learning_rate": 2.978499575710797e-07,
      "loss": 0.4037,
      "step": 5944
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4624218940734863,
      "learning_rate": 2.9702342590465674e-07,
      "loss": 0.4417,
      "step": 5945
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3692476749420166,
      "learning_rate": 2.9619800753508065e-07,
      "loss": 0.4801,
      "step": 5946
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.210471272468567,
      "learning_rate": 2.95373702657748e-07,
      "loss": 0.4857,
      "step": 5947
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1150034666061401,
      "learning_rate": 2.9455051146778766e-07,
      "loss": 0.5217,
      "step": 5948
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.256422758102417,
      "learning_rate": 2.9372843416006933e-07,
      "loss": 0.4879,
      "step": 5949
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0754444599151611,
      "learning_rate": 2.929074709291946e-07,
      "loss": 0.4355,
      "step": 5950
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4083384275436401,
      "learning_rate": 2.920876219695029e-07,
      "loss": 0.4572,
      "step": 5951
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5803221464157104,
      "learning_rate": 2.9126888747507296e-07,
      "loss": 0.5031,
      "step": 5952
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.9979984760284424,
      "learning_rate": 2.9045126763971475e-07,
      "loss": 0.4219,
      "step": 5953
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3697313070297241,
      "learning_rate": 2.8963476265697786e-07,
      "loss": 0.4567,
      "step": 5954
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1235144138336182,
      "learning_rate": 2.888193727201466e-07,
      "loss": 0.4441,
      "step": 5955
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.1492624282836914,
      "learning_rate": 2.8800509802224044e-07,
      "loss": 0.3935,
      "step": 5956
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.525577425956726,
      "learning_rate": 2.8719193875601794e-07,
      "loss": 0.4402,
      "step": 5957
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.213289737701416,
      "learning_rate": 2.86379895113969e-07,
      "loss": 0.4578,
      "step": 5958
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9284842014312744,
      "learning_rate": 2.855689672883238e-07,
      "loss": 0.4406,
      "step": 5959
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.0312604904174805,
      "learning_rate": 2.8475915547104606e-07,
      "loss": 0.4453,
      "step": 5960
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.0502240657806396,
      "learning_rate": 2.839504598538362e-07,
      "loss": 0.4258,
      "step": 5961
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4779547452926636,
      "learning_rate": 2.8314288062812946e-07,
      "loss": 0.4743,
      "step": 5962
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.1737442016601562,
      "learning_rate": 2.8233641798509794e-07,
      "loss": 0.4851,
      "step": 5963
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.285327672958374,
      "learning_rate": 2.815310721156489e-07,
      "loss": 0.4744,
      "step": 5964
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1112653017044067,
      "learning_rate": 2.807268432104243e-07,
      "loss": 0.475,
      "step": 5965
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4873539209365845,
      "learning_rate": 2.7992373145980465e-07,
      "loss": 0.4259,
      "step": 5966
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.964503824710846,
      "learning_rate": 2.7912173705390235e-07,
      "loss": 0.4299,
      "step": 5967
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2856361865997314,
      "learning_rate": 2.78320860182566e-07,
      "loss": 0.4988,
      "step": 5968
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.751641035079956,
      "learning_rate": 2.775211010353834e-07,
      "loss": 0.5006,
      "step": 5969
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4690555334091187,
      "learning_rate": 2.7672245980167214e-07,
      "loss": 0.5016,
      "step": 5970
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3005669116973877,
      "learning_rate": 2.759249366704897e-07,
      "loss": 0.4783,
      "step": 5971
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5226163864135742,
      "learning_rate": 2.7512853183062626e-07,
      "loss": 0.456,
      "step": 5972
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4761091470718384,
      "learning_rate": 2.743332454706088e-07,
      "loss": 0.4365,
      "step": 5973
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1606816053390503,
      "learning_rate": 2.735390777786989e-07,
      "loss": 0.405,
      "step": 5974
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.202021837234497,
      "learning_rate": 2.7274602894289235e-07,
      "loss": 0.4623,
      "step": 5975
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1625624895095825,
      "learning_rate": 2.719540991509223e-07,
      "loss": 0.4619,
      "step": 5976
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2029156684875488,
      "learning_rate": 2.7116328859025487e-07,
      "loss": 0.446,
      "step": 5977
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.873461127281189,
      "learning_rate": 2.703735974480931e-07,
      "loss": 0.4508,
      "step": 5978
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1972665786743164,
      "learning_rate": 2.695850259113736e-07,
      "loss": 0.4102,
      "step": 5979
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.248400330543518,
      "learning_rate": 2.6879757416676756e-07,
      "loss": 0.4941,
      "step": 5980
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.250422954559326,
      "learning_rate": 2.680112424006831e-07,
      "loss": 0.4883,
      "step": 5981
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3513883352279663,
      "learning_rate": 2.672260307992608e-07,
      "loss": 0.494,
      "step": 5982
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1863429546356201,
      "learning_rate": 2.664419395483786e-07,
      "loss": 0.4331,
      "step": 5983
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3468233346939087,
      "learning_rate": 2.656589688336475e-07,
      "loss": 0.4766,
      "step": 5984
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.582083821296692,
      "learning_rate": 2.648771188404131e-07,
      "loss": 0.4184,
      "step": 5985
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8548680543899536,
      "learning_rate": 2.640963897537574e-07,
      "loss": 0.4603,
      "step": 5986
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.445406436920166,
      "learning_rate": 2.6331678175849474e-07,
      "loss": 0.4638,
      "step": 5987
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.186980962753296,
      "learning_rate": 2.6253829503917694e-07,
      "loss": 0.4398,
      "step": 5988
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8827097415924072,
      "learning_rate": 2.6176092978008713e-07,
      "loss": 0.4534,
      "step": 5989
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4066494703292847,
      "learning_rate": 2.6098468616524544e-07,
      "loss": 0.4699,
      "step": 5990
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7276651859283447,
      "learning_rate": 2.602095643784058e-07,
      "loss": 0.4198,
      "step": 5991
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6578567028045654,
      "learning_rate": 2.5943556460305495e-07,
      "loss": 0.4762,
      "step": 5992
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2667131423950195,
      "learning_rate": 2.586626870224174e-07,
      "loss": 0.4173,
      "step": 5993
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.486224889755249,
      "learning_rate": 2.578909318194489e-07,
      "loss": 0.4711,
      "step": 5994
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.487863540649414,
      "learning_rate": 2.5712029917684014e-07,
      "loss": 0.4473,
      "step": 5995
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.1780426502227783,
      "learning_rate": 2.563507892770184e-07,
      "loss": 0.4893,
      "step": 5996
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.5025408267974854,
      "learning_rate": 2.5558240230214194e-07,
      "loss": 0.4223,
      "step": 5997
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4095940589904785,
      "learning_rate": 2.5481513843410576e-07,
      "loss": 0.4678,
      "step": 5998
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6315170526504517,
      "learning_rate": 2.5404899785453685e-07,
      "loss": 0.4434,
      "step": 5999
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2144560813903809,
      "learning_rate": 2.5328398074479844e-07,
      "loss": 0.3817,
      "step": 6000
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1517183780670166,
      "learning_rate": 2.525200872859856e-07,
      "loss": 0.4638,
      "step": 6001
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3172330856323242,
      "learning_rate": 2.517573176589294e-07,
      "loss": 0.5226,
      "step": 6002
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2696784734725952,
      "learning_rate": 2.509956720441936e-07,
      "loss": 0.4945,
      "step": 6003
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.264461636543274,
      "learning_rate": 2.502351506220757e-07,
      "loss": 0.4501,
      "step": 6004
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3264585733413696,
      "learning_rate": 2.4947575357260945e-07,
      "loss": 0.4832,
      "step": 6005
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.124776840209961,
      "learning_rate": 2.487174810755577e-07,
      "loss": 0.6488,
      "step": 6006
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1594576835632324,
      "learning_rate": 2.47960333310423e-07,
      "loss": 0.5264,
      "step": 6007
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2369812726974487,
      "learning_rate": 2.472043104564381e-07,
      "loss": 0.4441,
      "step": 6008
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2280009984970093,
      "learning_rate": 2.4644941269256873e-07,
      "loss": 0.4656,
      "step": 6009
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1753489971160889,
      "learning_rate": 2.4569564019751633e-07,
      "loss": 0.6974,
      "step": 6010
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.831458568572998,
      "learning_rate": 2.449429931497155e-07,
      "loss": 0.4876,
      "step": 6011
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2447655200958252,
      "learning_rate": 2.441914717273342e-07,
      "loss": 0.4013,
      "step": 6012
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2268568277359009,
      "learning_rate": 2.4344107610827283e-07,
      "loss": 0.4978,
      "step": 6013
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2252120971679688,
      "learning_rate": 2.4269180647016776e-07,
      "loss": 0.4352,
      "step": 6014
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.229393482208252,
      "learning_rate": 2.419436629903865e-07,
      "loss": 0.476,
      "step": 6015
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2993886470794678,
      "learning_rate": 2.411966458460319e-07,
      "loss": 0.4038,
      "step": 6016
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0958844423294067,
      "learning_rate": 2.404507552139379e-07,
      "loss": 0.4508,
      "step": 6017
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1384344100952148,
      "learning_rate": 2.397059912706734e-07,
      "loss": 0.6948,
      "step": 6018
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2818647623062134,
      "learning_rate": 2.389623541925407e-07,
      "loss": 0.4168,
      "step": 6019
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.441804051399231,
      "learning_rate": 2.3821984415557498e-07,
      "loss": 0.5174,
      "step": 6020
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2814394235610962,
      "learning_rate": 2.3747846133554353e-07,
      "loss": 0.4203,
      "step": 6021
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7503705024719238,
      "learning_rate": 2.367382059079487e-07,
      "loss": 0.497,
      "step": 6022
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2961612939834595,
      "learning_rate": 2.359990780480248e-07,
      "loss": 0.4734,
      "step": 6023
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.274132490158081,
      "learning_rate": 2.352610779307396e-07,
      "loss": 0.4298,
      "step": 6024
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2728633880615234,
      "learning_rate": 2.345242057307928e-07,
      "loss": 0.4305,
      "step": 6025
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3227558135986328,
      "learning_rate": 2.337884616226199e-07,
      "loss": 0.4925,
      "step": 6026
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2085607051849365,
      "learning_rate": 2.3305384578038547e-07,
      "loss": 0.4263,
      "step": 6027
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0263545513153076,
      "learning_rate": 2.323203583779904e-07,
      "loss": 0.4882,
      "step": 6028
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1632126569747925,
      "learning_rate": 2.3158799958906741e-07,
      "loss": 0.4636,
      "step": 6029
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.265045166015625,
      "learning_rate": 2.3085676958698012e-07,
      "loss": 0.437,
      "step": 6030
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1827208995819092,
      "learning_rate": 2.301266685448289e-07,
      "loss": 0.401,
      "step": 6031
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2593597173690796,
      "learning_rate": 2.2939769663544275e-07,
      "loss": 0.489,
      "step": 6032
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3579187393188477,
      "learning_rate": 2.2866985403138532e-07,
      "loss": 0.413,
      "step": 6033
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.309190034866333,
      "learning_rate": 2.2794314090495372e-07,
      "loss": 0.4445,
      "step": 6034
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4877220392227173,
      "learning_rate": 2.2721755742817542e-07,
      "loss": 0.4378,
      "step": 6035
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6189922094345093,
      "learning_rate": 2.2649310377281353e-07,
      "loss": 0.5206,
      "step": 6036
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5277750492095947,
      "learning_rate": 2.2576978011035978e-07,
      "loss": 0.4765,
      "step": 6037
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1516696214675903,
      "learning_rate": 2.250475866120433e-07,
      "loss": 0.4637,
      "step": 6038
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.1236817836761475,
      "learning_rate": 2.243265234488201e-07,
      "loss": 0.478,
      "step": 6039
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5226006507873535,
      "learning_rate": 2.236065907913837e-07,
      "loss": 0.4868,
      "step": 6040
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.1128735542297363,
      "learning_rate": 2.2288778881015716e-07,
      "loss": 0.3894,
      "step": 6041
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1108629703521729,
      "learning_rate": 2.2217011767529605e-07,
      "loss": 0.5119,
      "step": 6042
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1568540334701538,
      "learning_rate": 2.2145357755669006e-07,
      "loss": 0.4673,
      "step": 6043
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.570941686630249,
      "learning_rate": 2.2073816862395793e-07,
      "loss": 0.5048,
      "step": 6044
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9562433958053589,
      "learning_rate": 2.2002389104645472e-07,
      "loss": 0.4331,
      "step": 6045
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3756998777389526,
      "learning_rate": 2.1931074499326355e-07,
      "loss": 0.4251,
      "step": 6046
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9048160314559937,
      "learning_rate": 2.1859873063320214e-07,
      "loss": 0.5237,
      "step": 6047
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5866557359695435,
      "learning_rate": 2.1788784813482012e-07,
      "loss": 0.4851,
      "step": 6048
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1328716278076172,
      "learning_rate": 2.171780976663984e-07,
      "loss": 0.4293,
      "step": 6049
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1473333835601807,
      "learning_rate": 2.1646947939595153e-07,
      "loss": 0.501,
      "step": 6050
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.264647364616394,
      "learning_rate": 2.1576199349122307e-07,
      "loss": 0.4198,
      "step": 6051
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4610036611557007,
      "learning_rate": 2.150556401196907e-07,
      "loss": 0.4486,
      "step": 6052
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.252348780632019,
      "learning_rate": 2.1435041944856517e-07,
      "loss": 0.4424,
      "step": 6053
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1536823511123657,
      "learning_rate": 2.1364633164478622e-07,
      "loss": 0.4145,
      "step": 6054
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.312483787536621,
      "learning_rate": 2.129433768750272e-07,
      "loss": 0.4028,
      "step": 6055
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0974842309951782,
      "learning_rate": 2.1224155530569223e-07,
      "loss": 0.4156,
      "step": 6056
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.507382869720459,
      "learning_rate": 2.11540867102919e-07,
      "loss": 0.476,
      "step": 6057
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3522597551345825,
      "learning_rate": 2.1084131243257422e-07,
      "loss": 0.3898,
      "step": 6058
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4597619771957397,
      "learning_rate": 2.1014289146025823e-07,
      "loss": 0.4769,
      "step": 6059
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3411650657653809,
      "learning_rate": 2.0944560435130268e-07,
      "loss": 0.4356,
      "step": 6060
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2084364891052246,
      "learning_rate": 2.0874945127076994e-07,
      "loss": 0.5355,
      "step": 6061
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.469341516494751,
      "learning_rate": 2.0805443238345546e-07,
      "loss": 0.4783,
      "step": 6062
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6711598634719849,
      "learning_rate": 2.0736054785388536e-07,
      "loss": 0.4875,
      "step": 6063
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.306560754776001,
      "learning_rate": 2.0666779784631608e-07,
      "loss": 0.4892,
      "step": 6064
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.283546805381775,
      "learning_rate": 2.059761825247375e-07,
      "loss": 0.41,
      "step": 6065
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2284764051437378,
      "learning_rate": 2.052857020528698e-07,
      "loss": 0.7231,
      "step": 6066
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.5569164752960205,
      "learning_rate": 2.0459635659416498e-07,
      "loss": 0.4087,
      "step": 6067
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.122488021850586,
      "learning_rate": 2.0390814631180588e-07,
      "loss": 0.4673,
      "step": 6068
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3191606998443604,
      "learning_rate": 2.0322107136870662e-07,
      "loss": 0.4828,
      "step": 6069
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3091434240341187,
      "learning_rate": 2.0253513192751374e-07,
      "loss": 0.5201,
      "step": 6070
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4415686130523682,
      "learning_rate": 2.0185032815060235e-07,
      "loss": 0.3981,
      "step": 6071
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2570841312408447,
      "learning_rate": 2.0116666020008168e-07,
      "loss": 0.4832,
      "step": 6072
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1462782621383667,
      "learning_rate": 2.0048412823778952e-07,
      "loss": 0.4145,
      "step": 6073
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4010803699493408,
      "learning_rate": 1.9980273242529825e-07,
      "loss": 0.4538,
      "step": 6074
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7068161964416504,
      "learning_rate": 1.9912247292390774e-07,
      "loss": 0.4645,
      "step": 6075
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1893781423568726,
      "learning_rate": 1.9844334989464975e-07,
      "loss": 0.472,
      "step": 6076
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2382400035858154,
      "learning_rate": 1.9776536349828846e-07,
      "loss": 0.5521,
      "step": 6077
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4394358396530151,
      "learning_rate": 1.9708851389531714e-07,
      "loss": 0.388,
      "step": 6078
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1491055488586426,
      "learning_rate": 1.9641280124596153e-07,
      "loss": 0.7152,
      "step": 6079
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2427939176559448,
      "learning_rate": 1.9573822571017643e-07,
      "loss": 0.4331,
      "step": 6080
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3097131252288818,
      "learning_rate": 1.9506478744765022e-07,
      "loss": 0.4352,
      "step": 6081
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9499175548553467,
      "learning_rate": 1.9439248661779987e-07,
      "loss": 0.6726,
      "step": 6082
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4249165058135986,
      "learning_rate": 1.937213233797719e-07,
      "loss": 0.4862,
      "step": 6083
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4433823823928833,
      "learning_rate": 1.9305129789244648e-07,
      "loss": 0.5084,
      "step": 6084
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0947567224502563,
      "learning_rate": 1.9238241031443394e-07,
      "loss": 0.6994,
      "step": 6085
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5262807607650757,
      "learning_rate": 1.9171466080407374e-07,
      "loss": 0.4182,
      "step": 6086
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3022428750991821,
      "learning_rate": 1.9104804951943723e-07,
      "loss": 0.4938,
      "step": 6087
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.382798671722412,
      "learning_rate": 1.9038257661832483e-07,
      "loss": 0.4845,
      "step": 6088
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.4899301528930664,
      "learning_rate": 1.8971824225826996e-07,
      "loss": 0.4405,
      "step": 6089
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7216525077819824,
      "learning_rate": 1.8905504659653296e-07,
      "loss": 0.4654,
      "step": 6090
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.654536247253418,
      "learning_rate": 1.8839298979010823e-07,
      "loss": 0.5565,
      "step": 6091
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.349337100982666,
      "learning_rate": 1.8773207199571875e-07,
      "loss": 0.4029,
      "step": 6092
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2540053129196167,
      "learning_rate": 1.8707229336981768e-07,
      "loss": 0.4646,
      "step": 6093
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2075157165527344,
      "learning_rate": 1.8641365406858958e-07,
      "loss": 0.3838,
      "step": 6094
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2554380893707275,
      "learning_rate": 1.8575615424794747e-07,
      "loss": 0.5256,
      "step": 6095
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2778546810150146,
      "learning_rate": 1.8509979406353796e-07,
      "loss": 0.4382,
      "step": 6096
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2603373527526855,
      "learning_rate": 1.8444457367073344e-07,
      "loss": 0.345,
      "step": 6097
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0520542860031128,
      "learning_rate": 1.8379049322464094e-07,
      "loss": 0.4534,
      "step": 6098
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2929731607437134,
      "learning_rate": 1.831375528800944e-07,
      "loss": 0.4235,
      "step": 6099
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.215554118156433,
      "learning_rate": 1.8248575279165847e-07,
      "loss": 0.7071,
      "step": 6100
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.024501085281372,
      "learning_rate": 1.8183509311363035e-07,
      "loss": 0.5189,
      "step": 6101
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2903141975402832,
      "learning_rate": 1.8118557400003288e-07,
      "loss": 0.4724,
      "step": 6102
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6264675855636597,
      "learning_rate": 1.8053719560462312e-07,
      "loss": 0.4604,
      "step": 6103
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1820085048675537,
      "learning_rate": 1.798899580808855e-07,
      "loss": 0.3864,
      "step": 6104
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.1849818229675293,
      "learning_rate": 1.7924386158203576e-07,
      "loss": 0.4919,
      "step": 6105
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.578102469444275,
      "learning_rate": 1.7859890626101828e-07,
      "loss": 0.4304,
      "step": 6106
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3406273126602173,
      "learning_rate": 1.7795509227050868e-07,
      "loss": 0.489,
      "step": 6107
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.390621542930603,
      "learning_rate": 1.7731241976291226e-07,
      "loss": 0.4259,
      "step": 6108
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2976609468460083,
      "learning_rate": 1.7667088889036178e-07,
      "loss": 0.4638,
      "step": 6109
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2892553806304932,
      "learning_rate": 1.7603049980472352e-07,
      "loss": 0.4472,
      "step": 6110
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2571978569030762,
      "learning_rate": 1.7539125265759016e-07,
      "loss": 0.4291,
      "step": 6111
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1674872636795044,
      "learning_rate": 1.747531476002856e-07,
      "loss": 0.3848,
      "step": 6112
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1574965715408325,
      "learning_rate": 1.7411618478386406e-07,
      "loss": 0.4633,
      "step": 6113
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6621285676956177,
      "learning_rate": 1.7348036435910765e-07,
      "loss": 0.5255,
      "step": 6114
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.250604271888733,
      "learning_rate": 1.7284568647652934e-07,
      "loss": 0.4164,
      "step": 6115
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.275122046470642,
      "learning_rate": 1.7221215128637003e-07,
      "loss": 0.5634,
      "step": 6116
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.957638740539551,
      "learning_rate": 1.715797589386031e-07,
      "loss": 0.4559,
      "step": 6117
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1257004737854004,
      "learning_rate": 1.709485095829283e-07,
      "loss": 0.4605,
      "step": 6118
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5257878303527832,
      "learning_rate": 1.7031840336877603e-07,
      "loss": 0.374,
      "step": 6119
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1761868000030518,
      "learning_rate": 1.6968944044530755e-07,
      "loss": 0.421,
      "step": 6120
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1340912580490112,
      "learning_rate": 1.6906162096141043e-07,
      "loss": 0.4665,
      "step": 6121
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2238649129867554,
      "learning_rate": 1.6843494506570413e-07,
      "loss": 0.4167,
      "step": 6122
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.326644778251648,
      "learning_rate": 1.6780941290653662e-07,
      "loss": 0.4439,
      "step": 6123
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5084179639816284,
      "learning_rate": 1.6718502463198338e-07,
      "loss": 0.4487,
      "step": 6124
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.358247995376587,
      "learning_rate": 1.6656178038985282e-07,
      "loss": 0.4224,
      "step": 6125
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4236811399459839,
      "learning_rate": 1.6593968032767916e-07,
      "loss": 0.4818,
      "step": 6126
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.1606006622314453,
      "learning_rate": 1.6531872459272735e-07,
      "loss": 0.4627,
      "step": 6127
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1783348321914673,
      "learning_rate": 1.6469891333199095e-07,
      "loss": 0.4295,
      "step": 6128
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2242426872253418,
      "learning_rate": 1.640802466921926e-07,
      "loss": 0.4824,
      "step": 6129
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0654194355010986,
      "learning_rate": 1.6346272481978508e-07,
      "loss": 0.4432,
      "step": 6130
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2143652439117432,
      "learning_rate": 1.6284634786094765e-07,
      "loss": 0.7416,
      "step": 6131
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5351816415786743,
      "learning_rate": 1.6223111596159192e-07,
      "loss": 0.4452,
      "step": 6132
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2315117120742798,
      "learning_rate": 1.616170292673558e-07,
      "loss": 0.3995,
      "step": 6133
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4246922731399536,
      "learning_rate": 1.610040879236069e-07,
      "loss": 0.4377,
      "step": 6134
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2317944765090942,
      "learning_rate": 1.6039229207544194e-07,
      "loss": 0.3653,
      "step": 6135
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7556759119033813,
      "learning_rate": 1.597816418676862e-07,
      "loss": 0.5097,
      "step": 6136
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.213881492614746,
      "learning_rate": 1.59172137444894e-07,
      "loss": 0.4234,
      "step": 6137
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2744053602218628,
      "learning_rate": 1.5856377895134779e-07,
      "loss": 0.5075,
      "step": 6138
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2660613059997559,
      "learning_rate": 1.5795656653106007e-07,
      "loss": 0.4535,
      "step": 6139
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.19222891330719,
      "learning_rate": 1.5735050032777033e-07,
      "loss": 0.4541,
      "step": 6140
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5116840600967407,
      "learning_rate": 1.5674558048494826e-07,
      "loss": 0.4205,
      "step": 6141
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8097851276397705,
      "learning_rate": 1.5614180714579153e-07,
      "loss": 0.4325,
      "step": 6142
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4268007278442383,
      "learning_rate": 1.5553918045322636e-07,
      "loss": 0.3993,
      "step": 6143
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2927980422973633,
      "learning_rate": 1.5493770054990753e-07,
      "loss": 0.4826,
      "step": 6144
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3020062446594238,
      "learning_rate": 1.5433736757821838e-07,
      "loss": 0.4533,
      "step": 6145
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5377439260482788,
      "learning_rate": 1.5373818168027076e-07,
      "loss": 0.4604,
      "step": 6146
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.9538934230804443,
      "learning_rate": 1.5314014299790515e-07,
      "loss": 0.45,
      "step": 6147
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.285269021987915,
      "learning_rate": 1.525432516726899e-07,
      "loss": 0.4325,
      "step": 6148
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.337449312210083,
      "learning_rate": 1.5194750784592317e-07,
      "loss": 0.4431,
      "step": 6149
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3561484813690186,
      "learning_rate": 1.513529116586293e-07,
      "loss": 0.4585,
      "step": 6150
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2511239051818848,
      "learning_rate": 1.5075946325156242e-07,
      "loss": 0.4661,
      "step": 6151
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.224979281425476,
      "learning_rate": 1.501671627652057e-07,
      "loss": 0.4363,
      "step": 6152
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7596566677093506,
      "learning_rate": 1.4957601033976755e-07,
      "loss": 0.5123,
      "step": 6153
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.298445463180542,
      "learning_rate": 1.4898600611518876e-07,
      "loss": 0.4271,
      "step": 6154
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2687811851501465,
      "learning_rate": 1.4839715023113487e-07,
      "loss": 0.4874,
      "step": 6155
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3689980506896973,
      "learning_rate": 1.478094428270016e-07,
      "loss": 0.4764,
      "step": 6156
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0909415483474731,
      "learning_rate": 1.4722288404191098e-07,
      "loss": 0.3719,
      "step": 6157
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1846619844436646,
      "learning_rate": 1.4663747401471584e-07,
      "loss": 0.4491,
      "step": 6158
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1982311010360718,
      "learning_rate": 1.4605321288399365e-07,
      "loss": 0.4253,
      "step": 6159
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4738526344299316,
      "learning_rate": 1.454701007880538e-07,
      "loss": 0.4222,
      "step": 6160
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.091627836227417,
      "learning_rate": 1.4488813786492974e-07,
      "loss": 0.4422,
      "step": 6161
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.625881314277649,
      "learning_rate": 1.443073242523846e-07,
      "loss": 0.4132,
      "step": 6162
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1542006731033325,
      "learning_rate": 1.4372766008791174e-07,
      "loss": 0.4115,
      "step": 6163
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3168131113052368,
      "learning_rate": 1.431491455087286e-07,
      "loss": 0.4969,
      "step": 6164
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3919569253921509,
      "learning_rate": 1.4257178065178233e-07,
      "loss": 0.4505,
      "step": 6165
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1946887969970703,
      "learning_rate": 1.419955656537486e-07,
      "loss": 0.4575,
      "step": 6166
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5941849946975708,
      "learning_rate": 1.4142050065102885e-07,
      "loss": 0.5284,
      "step": 6167
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3454581499099731,
      "learning_rate": 1.4084658577975417e-07,
      "loss": 0.3893,
      "step": 6168
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5534276962280273,
      "learning_rate": 1.402738211757826e-07,
      "loss": 0.3562,
      "step": 6169
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1727811098098755,
      "learning_rate": 1.3970220697470005e-07,
      "loss": 0.4121,
      "step": 6170
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.238834261894226,
      "learning_rate": 1.3913174331181946e-07,
      "loss": 0.459,
      "step": 6171
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1024794578552246,
      "learning_rate": 1.3856243032218274e-07,
      "loss": 0.4191,
      "step": 6172
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.009436845779419,
      "learning_rate": 1.3799426814055827e-07,
      "loss": 0.4676,
      "step": 6173
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0770608186721802,
      "learning_rate": 1.3742725690144233e-07,
      "loss": 0.4424,
      "step": 6174
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2015584707260132,
      "learning_rate": 1.3686139673905973e-07,
      "loss": 0.5091,
      "step": 6175
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.736302375793457,
      "learning_rate": 1.3629668778736117e-07,
      "loss": 0.4737,
      "step": 6176
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7372876405715942,
      "learning_rate": 1.3573313018002466e-07,
      "loss": 0.4324,
      "step": 6177
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9224064350128174,
      "learning_rate": 1.3517072405045796e-07,
      "loss": 0.4366,
      "step": 6178
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.437030553817749,
      "learning_rate": 1.3460946953179343e-07,
      "loss": 0.4233,
      "step": 6179
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.880394458770752,
      "learning_rate": 1.3404936675689428e-07,
      "loss": 0.4559,
      "step": 6180
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1873236894607544,
      "learning_rate": 1.3349041585834664e-07,
      "loss": 0.4256,
      "step": 6181
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1697320938110352,
      "learning_rate": 1.3293261696846804e-07,
      "loss": 0.4638,
      "step": 6182
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4743092060089111,
      "learning_rate": 1.3237597021930116e-07,
      "loss": 0.4495,
      "step": 6183
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2101274728775024,
      "learning_rate": 1.3182047574261557e-07,
      "loss": 0.4354,
      "step": 6184
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1060141324996948,
      "learning_rate": 1.312661336699106e-07,
      "loss": 0.4227,
      "step": 6185
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4082225561141968,
      "learning_rate": 1.3071294413240954e-07,
      "loss": 0.509,
      "step": 6186
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0730904340744019,
      "learning_rate": 1.3016090726106546e-07,
      "loss": 0.3866,
      "step": 6187
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.556258201599121,
      "learning_rate": 1.2961002318655603e-07,
      "loss": 0.4876,
      "step": 6188
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3020365238189697,
      "learning_rate": 1.2906029203928917e-07,
      "loss": 0.4729,
      "step": 6189
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3689016103744507,
      "learning_rate": 1.2851171394939743e-07,
      "loss": 0.48,
      "step": 6190
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2837653160095215,
      "learning_rate": 1.2796428904674085e-07,
      "loss": 0.5112,
      "step": 6191
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3132967948913574,
      "learning_rate": 1.27418017460908e-07,
      "loss": 0.4393,
      "step": 6192
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4075515270233154,
      "learning_rate": 1.268728993212115e-07,
      "loss": 0.4503,
      "step": 6193
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5306932926177979,
      "learning_rate": 1.2632893475669428e-07,
      "loss": 0.4283,
      "step": 6194
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1301289796829224,
      "learning_rate": 1.2578612389612333e-07,
      "loss": 0.4892,
      "step": 6195
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3910197019577026,
      "learning_rate": 1.2524446686799418e-07,
      "loss": 0.5587,
      "step": 6196
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.4750099182128906,
      "learning_rate": 1.2470396380052986e-07,
      "loss": 0.4552,
      "step": 6197
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.601370930671692,
      "learning_rate": 1.2416461482167742e-07,
      "loss": 0.468,
      "step": 6198
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.242173433303833,
      "learning_rate": 1.2362642005911418e-07,
      "loss": 0.3962,
      "step": 6199
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.254649043083191,
      "learning_rate": 1.2308937964024105e-07,
      "loss": 0.4899,
      "step": 6200
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.098645806312561,
      "learning_rate": 1.225534936921885e-07,
      "loss": 0.4364,
      "step": 6201
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3453888893127441,
      "learning_rate": 1.2201876234181232e-07,
      "loss": 0.4902,
      "step": 6202
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.560877799987793,
      "learning_rate": 1.2148518571569346e-07,
      "loss": 0.5237,
      "step": 6203
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.305488109588623,
      "learning_rate": 1.2095276394014254e-07,
      "loss": 0.4426,
      "step": 6204
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2718048095703125,
      "learning_rate": 1.2042149714119488e-07,
      "loss": 0.5046,
      "step": 6205
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1050286293029785,
      "learning_rate": 1.1989138544461375e-07,
      "loss": 0.5401,
      "step": 6206
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1005947589874268,
      "learning_rate": 1.1936242897588656e-07,
      "loss": 0.4229,
      "step": 6207
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4484282732009888,
      "learning_rate": 1.1883462786022982e-07,
      "loss": 0.4455,
      "step": 6208
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2275344133377075,
      "learning_rate": 1.1830798222258577e-07,
      "loss": 0.5345,
      "step": 6209
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.084150552749634,
      "learning_rate": 1.1778249218762194e-07,
      "loss": 0.4434,
      "step": 6210
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2760833501815796,
      "learning_rate": 1.1725815787973438e-07,
      "loss": 0.4779,
      "step": 6211
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.4146416187286377,
      "learning_rate": 1.1673497942304324e-07,
      "loss": 0.4907,
      "step": 6212
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.967536687850952,
      "learning_rate": 1.1621295694139723e-07,
      "loss": 0.4448,
      "step": 6213
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.84062659740448,
      "learning_rate": 1.1569209055837027e-07,
      "loss": 0.5533,
      "step": 6214
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6099687814712524,
      "learning_rate": 1.1517238039726153e-07,
      "loss": 0.4334,
      "step": 6215
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1588470935821533,
      "learning_rate": 1.146538265810998e-07,
      "loss": 0.6907,
      "step": 6216
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4531892538070679,
      "learning_rate": 1.1413642923263635e-07,
      "loss": 0.4189,
      "step": 6217
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0738111734390259,
      "learning_rate": 1.1362018847435097e-07,
      "loss": 0.4011,
      "step": 6218
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2512660026550293,
      "learning_rate": 1.1310510442844924e-07,
      "loss": 0.4174,
      "step": 6219
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3367427587509155,
      "learning_rate": 1.1259117721686252e-07,
      "loss": 0.7176,
      "step": 6220
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3731184005737305,
      "learning_rate": 1.1207840696124905e-07,
      "loss": 0.4723,
      "step": 6221
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1961472034454346,
      "learning_rate": 1.1156679378299174e-07,
      "loss": 0.6657,
      "step": 6222
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7881290912628174,
      "learning_rate": 1.1105633780320146e-07,
      "loss": 0.4768,
      "step": 6223
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1898022890090942,
      "learning_rate": 1.105470391427138e-07,
      "loss": 0.5105,
      "step": 6224
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5506694316864014,
      "learning_rate": 1.1003889792209122e-07,
      "loss": 0.5172,
      "step": 6225
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2464816570281982,
      "learning_rate": 1.0953191426162136e-07,
      "loss": 0.4197,
      "step": 6226
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.221421480178833,
      "learning_rate": 1.0902608828131822e-07,
      "loss": 0.4301,
      "step": 6227
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2780271768569946,
      "learning_rate": 1.085214201009227e-07,
      "loss": 0.3651,
      "step": 6228
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2030264139175415,
      "learning_rate": 1.0801790983989869e-07,
      "loss": 0.4437,
      "step": 6229
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2974909543991089,
      "learning_rate": 1.0751555761744026e-07,
      "loss": 0.4562,
      "step": 6230
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4097820520401,
      "learning_rate": 1.0701436355246452e-07,
      "loss": 0.5177,
      "step": 6231
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6491726636886597,
      "learning_rate": 1.0651432776361437e-07,
      "loss": 0.5218,
      "step": 6232
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2722300291061401,
      "learning_rate": 1.0601545036925953e-07,
      "loss": 0.4846,
      "step": 6233
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1929091215133667,
      "learning_rate": 1.05517731487495e-07,
      "loss": 0.4442,
      "step": 6234
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3017038106918335,
      "learning_rate": 1.0502117123614209e-07,
      "loss": 0.5116,
      "step": 6235
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3276389837265015,
      "learning_rate": 1.0452576973274675e-07,
      "loss": 0.4085,
      "step": 6236
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2674428224563599,
      "learning_rate": 1.0403152709458186e-07,
      "loss": 0.4638,
      "step": 6237
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.936220407485962,
      "learning_rate": 1.035384434386455e-07,
      "loss": 0.5128,
      "step": 6238
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1770895719528198,
      "learning_rate": 1.0304651888166039e-07,
      "loss": 0.4489,
      "step": 6239
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0476661920547485,
      "learning_rate": 1.0255575354007619e-07,
      "loss": 0.4295,
      "step": 6240
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.478857398033142,
      "learning_rate": 1.0206614753006827e-07,
      "loss": 0.4382,
      "step": 6241
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1064605712890625,
      "learning_rate": 1.0157770096753672e-07,
      "loss": 0.3884,
      "step": 6242
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.6798770427703857,
      "learning_rate": 1.0109041396810738e-07,
      "loss": 0.4846,
      "step": 6243
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.360503077507019,
      "learning_rate": 1.0060428664713129e-07,
      "loss": 0.4468,
      "step": 6244
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6536117792129517,
      "learning_rate": 1.0011931911968642e-07,
      "loss": 0.4862,
      "step": 6245
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.29509437084198,
      "learning_rate": 9.963551150057427e-08,
      "loss": 0.4894,
      "step": 6246
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.159639835357666,
      "learning_rate": 9.915286390432266e-08,
      "loss": 0.4109,
      "step": 6247
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3302152156829834,
      "learning_rate": 9.86713764451852e-08,
      "loss": 0.5345,
      "step": 6248
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.56699275970459,
      "learning_rate": 9.819104923714018e-08,
      "loss": 0.4622,
      "step": 6249
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.073800802230835,
      "learning_rate": 9.771188239389162e-08,
      "loss": 0.5567,
      "step": 6250
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2472081184387207,
      "learning_rate": 9.723387602886769e-08,
      "loss": 0.4445,
      "step": 6251
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1960840225219727,
      "learning_rate": 9.67570302552251e-08,
      "loss": 0.4419,
      "step": 6252
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.993818998336792,
      "learning_rate": 9.628134518584131e-08,
      "loss": 0.4138,
      "step": 6253
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5014328956604004,
      "learning_rate": 9.580682093332295e-08,
      "loss": 0.4488,
      "step": 6254
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4516539573669434,
      "learning_rate": 9.533345761000012e-08,
      "loss": 0.3919,
      "step": 6255
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2557713985443115,
      "learning_rate": 9.486125532792656e-08,
      "loss": 0.4342,
      "step": 6256
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3944579362869263,
      "learning_rate": 9.439021419888505e-08,
      "loss": 0.4599,
      "step": 6257
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.516564965248108,
      "learning_rate": 9.392033433437975e-08,
      "loss": 0.4401,
      "step": 6258
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1544735431671143,
      "learning_rate": 9.345161584564221e-08,
      "loss": 0.4279,
      "step": 6259
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3305288553237915,
      "learning_rate": 9.29840588436276e-08,
      "loss": 0.4521,
      "step": 6260
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3008249998092651,
      "learning_rate": 9.251766343901735e-08,
      "loss": 0.4433,
      "step": 6261
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3935608863830566,
      "learning_rate": 9.205242974221707e-08,
      "loss": 0.45,
      "step": 6262
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0151567459106445,
      "learning_rate": 9.158835786335806e-08,
      "loss": 0.6509,
      "step": 6263
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3789466619491577,
      "learning_rate": 9.112544791229638e-08,
      "loss": 0.4467,
      "step": 6264
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4854058027267456,
      "learning_rate": 9.066369999861213e-08,
      "loss": 0.4761,
      "step": 6265
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4212145805358887,
      "learning_rate": 9.020311423161177e-08,
      "loss": 0.4558,
      "step": 6266
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8683171272277832,
      "learning_rate": 8.974369072032584e-08,
      "loss": 0.4796,
      "step": 6267
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.150705337524414,
      "learning_rate": 8.928542957350961e-08,
      "loss": 0.7128,
      "step": 6268
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0425670146942139,
      "learning_rate": 8.882833089964349e-08,
      "loss": 0.4555,
      "step": 6269
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1731750965118408,
      "learning_rate": 8.83723948069326e-08,
      "loss": 0.4597,
      "step": 6270
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6974947452545166,
      "learning_rate": 8.791762140330784e-08,
      "loss": 0.4701,
      "step": 6271
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2732880115509033,
      "learning_rate": 8.746401079642252e-08,
      "loss": 0.5131,
      "step": 6272
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3858835697174072,
      "learning_rate": 8.701156309365688e-08,
      "loss": 0.4517,
      "step": 6273
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0665796995162964,
      "learning_rate": 8.656027840211578e-08,
      "loss": 0.4825,
      "step": 6274
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.339457392692566,
      "learning_rate": 8.611015682862711e-08,
      "loss": 0.4631,
      "step": 6275
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.389053225517273,
      "learning_rate": 8.566119847974508e-08,
      "loss": 0.4623,
      "step": 6276
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7013863325119019,
      "learning_rate": 8.521340346174689e-08,
      "loss": 0.5067,
      "step": 6277
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2577224969863892,
      "learning_rate": 8.47667718806372e-08,
      "loss": 0.4552,
      "step": 6278
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3795610666275024,
      "learning_rate": 8.432130384214143e-08,
      "loss": 0.5454,
      "step": 6279
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2567541599273682,
      "learning_rate": 8.387699945171245e-08,
      "loss": 0.4286,
      "step": 6280
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.253307819366455,
      "learning_rate": 8.343385881452726e-08,
      "loss": 0.4641,
      "step": 6281
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8830512762069702,
      "learning_rate": 8.29918820354858e-08,
      "loss": 0.4587,
      "step": 6282
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9357942342758179,
      "learning_rate": 8.255106921921497e-08,
      "loss": 0.3916,
      "step": 6283
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5446871519088745,
      "learning_rate": 8.211142047006293e-08,
      "loss": 0.4565,
      "step": 6284
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1932085752487183,
      "learning_rate": 8.167293589210534e-08,
      "loss": 0.4611,
      "step": 6285
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3052386045455933,
      "learning_rate": 8.12356155891414e-08,
      "loss": 0.4714,
      "step": 6286
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.0532641410827637,
      "learning_rate": 8.079945966469327e-08,
      "loss": 0.441,
      "step": 6287
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.301807165145874,
      "learning_rate": 8.03644682220095e-08,
      "loss": 0.5183,
      "step": 6288
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4763785600662231,
      "learning_rate": 7.993064136406103e-08,
      "loss": 0.4054,
      "step": 6289
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1773709058761597,
      "learning_rate": 7.949797919354518e-08,
      "loss": 0.4595,
      "step": 6290
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4586358070373535,
      "learning_rate": 7.906648181288168e-08,
      "loss": 0.3806,
      "step": 6291
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2388416528701782,
      "learning_rate": 7.863614932421548e-08,
      "loss": 0.4464,
      "step": 6292
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.8573715686798096,
      "learning_rate": 7.820698182941621e-08,
      "loss": 0.4034,
      "step": 6293
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5247457027435303,
      "learning_rate": 7.777897943007595e-08,
      "loss": 0.4395,
      "step": 6294
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5742689371109009,
      "learning_rate": 7.735214222751363e-08,
      "loss": 0.5289,
      "step": 6295
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2004671096801758,
      "learning_rate": 7.692647032276956e-08,
      "loss": 0.4159,
      "step": 6296
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4283455610275269,
      "learning_rate": 7.650196381661035e-08,
      "loss": 0.5,
      "step": 6297
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.383331537246704,
      "learning_rate": 7.607862280952616e-08,
      "loss": 0.4864,
      "step": 6298
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2958974838256836,
      "learning_rate": 7.565644740173073e-08,
      "loss": 0.4494,
      "step": 6299
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7633726596832275,
      "learning_rate": 7.523543769316188e-08,
      "loss": 0.5081,
      "step": 6300
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1810506582260132,
      "learning_rate": 7.481559378348158e-08,
      "loss": 0.4787,
      "step": 6301
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2896037101745605,
      "learning_rate": 7.439691577207697e-08,
      "loss": 0.4435,
      "step": 6302
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4946660995483398,
      "learning_rate": 7.397940375805712e-08,
      "loss": 0.4982,
      "step": 6303
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.200614094734192,
      "learning_rate": 7.356305784025686e-08,
      "loss": 0.4935,
      "step": 6304
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0408453941345215,
      "learning_rate": 7.314787811723456e-08,
      "loss": 0.3689,
      "step": 6305
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3481441736221313,
      "learning_rate": 7.273386468727162e-08,
      "loss": 0.4975,
      "step": 6306
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.882988452911377,
      "learning_rate": 7.232101764837407e-08,
      "loss": 0.4945,
      "step": 6307
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4166797399520874,
      "learning_rate": 7.190933709827208e-08,
      "loss": 0.4446,
      "step": 6308
    },
    {
      "epoch": 0.95,
      "grad_norm": 53.36237716674805,
      "learning_rate": 7.149882313441991e-08,
      "loss": 0.4408,
      "step": 6309
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0526978969573975,
      "learning_rate": 7.108947585399429e-08,
      "loss": 0.4629,
      "step": 6310
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.770702362060547,
      "learning_rate": 7.068129535389657e-08,
      "loss": 0.3745,
      "step": 6311
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1018600463867188,
      "learning_rate": 7.027428173075279e-08,
      "loss": 0.7185,
      "step": 6312
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7967737913131714,
      "learning_rate": 6.986843508091146e-08,
      "loss": 0.4636,
      "step": 6313
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3956245183944702,
      "learning_rate": 6.946375550044515e-08,
      "loss": 0.4462,
      "step": 6314
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2591108083724976,
      "learning_rate": 6.906024308515003e-08,
      "loss": 0.4019,
      "step": 6315
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.7687692642211914,
      "learning_rate": 6.865789793054745e-08,
      "loss": 0.4782,
      "step": 6316
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2495721578598022,
      "learning_rate": 6.825672013188012e-08,
      "loss": 0.3801,
      "step": 6317
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.323906421661377,
      "learning_rate": 6.78567097841154e-08,
      "loss": 0.4364,
      "step": 6318
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.2707324028015137,
      "learning_rate": 6.745786698194589e-08,
      "loss": 0.4528,
      "step": 6319
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.048696756362915,
      "learning_rate": 6.706019181978551e-08,
      "loss": 0.6871,
      "step": 6320
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3836708068847656,
      "learning_rate": 6.666368439177173e-08,
      "loss": 0.5003,
      "step": 6321
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7958729267120361,
      "learning_rate": 6.62683447917678e-08,
      "loss": 0.4534,
      "step": 6322
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0775270462036133,
      "learning_rate": 6.587417311335886e-08,
      "loss": 0.3812,
      "step": 6323
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0363035202026367,
      "learning_rate": 6.548116944985361e-08,
      "loss": 0.444,
      "step": 6324
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.163611650466919,
      "learning_rate": 6.50893338942843e-08,
      "loss": 0.4306,
      "step": 6325
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5371007919311523,
      "learning_rate": 6.469866653940782e-08,
      "loss": 0.4822,
      "step": 6326
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9552630186080933,
      "learning_rate": 6.430916747770299e-08,
      "loss": 0.5536,
      "step": 6327
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.78888738155365,
      "learning_rate": 6.392083680137329e-08,
      "loss": 0.4977,
      "step": 6328
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1821788549423218,
      "learning_rate": 6.353367460234406e-08,
      "loss": 0.4142,
      "step": 6329
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.206154227256775,
      "learning_rate": 6.314768097226586e-08,
      "loss": 0.4387,
      "step": 6330
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.461766242980957,
      "learning_rate": 6.27628560025112e-08,
      "loss": 0.4576,
      "step": 6331
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1557966470718384,
      "learning_rate": 6.237919978417717e-08,
      "loss": 0.4418,
      "step": 6332
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.195461630821228,
      "learning_rate": 6.199671240808336e-08,
      "loss": 0.4435,
      "step": 6333
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.398201823234558,
      "learning_rate": 6.161539396477234e-08,
      "loss": 0.4646,
      "step": 6334
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.090097188949585,
      "learning_rate": 6.123524454451079e-08,
      "loss": 0.4699,
      "step": 6335
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2488477230072021,
      "learning_rate": 6.085626423728897e-08,
      "loss": 0.4568,
      "step": 6336
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5976011753082275,
      "learning_rate": 6.047845313281842e-08,
      "loss": 0.502,
      "step": 6337
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.098710060119629,
      "learning_rate": 6.010181132053649e-08,
      "loss": 0.4904,
      "step": 6338
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5496398210525513,
      "learning_rate": 5.972633888960133e-08,
      "loss": 0.4549,
      "step": 6339
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5038987398147583,
      "learning_rate": 5.935203592889627e-08,
      "loss": 0.4968,
      "step": 6340
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.278618335723877,
      "learning_rate": 5.897890252702765e-08,
      "loss": 0.4331,
      "step": 6341
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.037022352218628,
      "learning_rate": 5.86069387723226e-08,
      "loss": 0.7248,
      "step": 6342
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2888376712799072,
      "learning_rate": 5.823614475283401e-08,
      "loss": 0.441,
      "step": 6343
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4202766418457031,
      "learning_rate": 5.786652055633668e-08,
      "loss": 0.433,
      "step": 6344
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1730644702911377,
      "learning_rate": 5.749806627032895e-08,
      "loss": 0.4718,
      "step": 6345
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.028319001197815,
      "learning_rate": 5.713078198203215e-08,
      "loss": 0.4416,
      "step": 6346
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6743897199630737,
      "learning_rate": 5.676466777838896e-08,
      "loss": 0.4706,
      "step": 6347
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3776814937591553,
      "learning_rate": 5.639972374606839e-08,
      "loss": 0.4996,
      "step": 6348
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.261216163635254,
      "learning_rate": 5.603594997145967e-08,
      "loss": 0.4874,
      "step": 6349
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.282099962234497,
      "learning_rate": 5.5673346540676684e-08,
      "loss": 0.4602,
      "step": 6350
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3415552377700806,
      "learning_rate": 5.531191353955412e-08,
      "loss": 0.4648,
      "step": 6351
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.0670278072357178,
      "learning_rate": 5.495165105365241e-08,
      "loss": 0.4517,
      "step": 6352
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3080847263336182,
      "learning_rate": 5.459255916825335e-08,
      "loss": 0.4311,
      "step": 6353
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.660029411315918,
      "learning_rate": 5.4234637968361704e-08,
      "loss": 0.4535,
      "step": 6354
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.423103928565979,
      "learning_rate": 5.387788753870471e-08,
      "loss": 0.4324,
      "step": 6355
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7883342504501343,
      "learning_rate": 5.3522307963733146e-08,
      "loss": 0.5247,
      "step": 6356
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4473998546600342,
      "learning_rate": 5.3167899327620784e-08,
      "loss": 0.4418,
      "step": 6357
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7969449758529663,
      "learning_rate": 5.281466171426386e-08,
      "loss": 0.4466,
      "step": 6358
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6992650032043457,
      "learning_rate": 5.246259520728103e-08,
      "loss": 0.5481,
      "step": 6359
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.33754563331604,
      "learning_rate": 5.211169989001397e-08,
      "loss": 0.42,
      "step": 6360
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.140332579612732,
      "learning_rate": 5.176197584552789e-08,
      "loss": 0.4368,
      "step": 6361
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.245640516281128,
      "learning_rate": 5.1413423156609355e-08,
      "loss": 0.4747,
      "step": 6362
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3632584810256958,
      "learning_rate": 5.106604190576902e-08,
      "loss": 0.4368,
      "step": 6363
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2714790105819702,
      "learning_rate": 5.07198321752389e-08,
      "loss": 0.4759,
      "step": 6364
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0768886804580688,
      "learning_rate": 5.03747940469751e-08,
      "loss": 0.399,
      "step": 6365
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.7144081592559814,
      "learning_rate": 5.003092760265504e-08,
      "loss": 0.529,
      "step": 6366
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.296773076057434,
      "learning_rate": 4.9688232923679744e-08,
      "loss": 0.4256,
      "step": 6367
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6028409004211426,
      "learning_rate": 4.9346710091172646e-08,
      "loss": 0.5059,
      "step": 6368
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4044525623321533,
      "learning_rate": 4.900635918597907e-08,
      "loss": 0.4182,
      "step": 6369
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7163218259811401,
      "learning_rate": 4.86671802886679e-08,
      "loss": 0.5152,
      "step": 6370
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.612950325012207,
      "learning_rate": 4.832917347952992e-08,
      "loss": 0.4915,
      "step": 6371
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4085925817489624,
      "learning_rate": 4.799233883857946e-08,
      "loss": 0.4628,
      "step": 6372
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3101989030838013,
      "learning_rate": 4.7656676445551054e-08,
      "loss": 0.4554,
      "step": 6373
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2524479627609253,
      "learning_rate": 4.7322186379904486e-08,
      "loss": 0.4012,
      "step": 6374
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0516471862792969,
      "learning_rate": 4.698886872082087e-08,
      "loss": 0.4409,
      "step": 6375
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1544685363769531,
      "learning_rate": 4.665672354720319e-08,
      "loss": 0.481,
      "step": 6376
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4073227643966675,
      "learning_rate": 4.632575093767744e-08,
      "loss": 0.4202,
      "step": 6377
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.47385835647583,
      "learning_rate": 4.5995950970592616e-08,
      "loss": 0.4466,
      "step": 6378
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9231927394866943,
      "learning_rate": 4.5667323724019055e-08,
      "loss": 0.4719,
      "step": 6379
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7641595602035522,
      "learning_rate": 4.533986927575007e-08,
      "loss": 0.5118,
      "step": 6380
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.240182876586914,
      "learning_rate": 4.501358770330144e-08,
      "loss": 0.3946,
      "step": 6381
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8270344734191895,
      "learning_rate": 4.468847908391083e-08,
      "loss": 0.4476,
      "step": 6382
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4127408266067505,
      "learning_rate": 4.436454349453834e-08,
      "loss": 0.4501,
      "step": 6383
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2941384315490723,
      "learning_rate": 4.4041781011867624e-08,
      "loss": 0.4639,
      "step": 6384
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5263066291809082,
      "learning_rate": 4.3720191712302016e-08,
      "loss": 0.4044,
      "step": 6385
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4513424634933472,
      "learning_rate": 4.3399775671970066e-08,
      "loss": 0.4258,
      "step": 6386
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4863282442092896,
      "learning_rate": 4.3080532966720546e-08,
      "loss": 0.4419,
      "step": 6387
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.317551851272583,
      "learning_rate": 4.276246367212522e-08,
      "loss": 0.5219,
      "step": 6388
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5166012048721313,
      "learning_rate": 4.244556786347831e-08,
      "loss": 0.5544,
      "step": 6389
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.308801293373108,
      "learning_rate": 4.2129845615794806e-08,
      "loss": 0.4617,
      "step": 6390
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2161593437194824,
      "learning_rate": 4.1815297003814923e-08,
      "loss": 0.4863,
      "step": 6391
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.468869686126709,
      "learning_rate": 4.150192210199799e-08,
      "loss": 0.4537,
      "step": 6392
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4816311597824097,
      "learning_rate": 4.1189720984526337e-08,
      "loss": 0.4968,
      "step": 6393
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8064178228378296,
      "learning_rate": 4.087869372530584e-08,
      "loss": 0.3915,
      "step": 6394
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.551382303237915,
      "learning_rate": 4.056884039796261e-08,
      "loss": 0.4645,
      "step": 6395
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4100602865219116,
      "learning_rate": 4.026016107584574e-08,
      "loss": 0.4591,
      "step": 6396
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3480697870254517,
      "learning_rate": 3.995265583202623e-08,
      "loss": 0.3767,
      "step": 6397
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4094678163528442,
      "learning_rate": 3.964632473929808e-08,
      "loss": 0.492,
      "step": 6398
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4811170101165771,
      "learning_rate": 3.9341167870176056e-08,
      "loss": 0.5151,
      "step": 6399
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2463675737380981,
      "learning_rate": 3.903718529689682e-08,
      "loss": 0.3912,
      "step": 6400
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.296769618988037,
      "learning_rate": 3.873437709142058e-08,
      "loss": 0.4744,
      "step": 6401
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4697611331939697,
      "learning_rate": 3.843274332542834e-08,
      "loss": 0.4442,
      "step": 6402
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2308117151260376,
      "learning_rate": 3.813228407032299e-08,
      "loss": 0.4084,
      "step": 6403
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3584986925125122,
      "learning_rate": 3.783299939722984e-08,
      "loss": 0.4092,
      "step": 6404
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4332027435302734,
      "learning_rate": 3.7534889376996676e-08,
      "loss": 0.4533,
      "step": 6405
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0118104219436646,
      "learning_rate": 3.723795408019204e-08,
      "loss": 0.6871,
      "step": 6406
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4148446321487427,
      "learning_rate": 3.6942193577106377e-08,
      "loss": 0.4646,
      "step": 6407
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.30073082447052,
      "learning_rate": 3.6647607937754235e-08,
      "loss": 0.4153,
      "step": 6408
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5682467222213745,
      "learning_rate": 3.635419723186984e-08,
      "loss": 0.3985,
      "step": 6409
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9581263065338135,
      "learning_rate": 3.606196152890928e-08,
      "loss": 0.4662,
      "step": 6410
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4996120929718018,
      "learning_rate": 3.5770900898051685e-08,
      "loss": 0.4819,
      "step": 6411
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6213451623916626,
      "learning_rate": 3.548101540819693e-08,
      "loss": 0.4974,
      "step": 6412
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.1731488704681396,
      "learning_rate": 3.5192305127967894e-08,
      "loss": 0.4272,
      "step": 6413
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.0140316486358643,
      "learning_rate": 3.4904770125707696e-08,
      "loss": 0.5684,
      "step": 6414
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0444865226745605,
      "learning_rate": 3.4618410469483e-08,
      "loss": 0.4306,
      "step": 6415
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.108139157295227,
      "learning_rate": 3.4333226227080685e-08,
      "loss": 0.4872,
      "step": 6416
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1546272039413452,
      "learning_rate": 3.40492174660112e-08,
      "loss": 0.6727,
      "step": 6417
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0731843709945679,
      "learning_rate": 3.376638425350409e-08,
      "loss": 0.728,
      "step": 6418
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4585076570510864,
      "learning_rate": 3.348472665651248e-08,
      "loss": 0.4606,
      "step": 6419
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.250848770141602,
      "learning_rate": 3.3204244741711913e-08,
      "loss": 0.4388,
      "step": 6420
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.1993119716644287,
      "learning_rate": 3.292493857549761e-08,
      "loss": 0.4237,
      "step": 6421
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.753817081451416,
      "learning_rate": 3.2646808223987804e-08,
      "loss": 0.4251,
      "step": 6422
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4928481578826904,
      "learning_rate": 3.236985375302148e-08,
      "loss": 0.4332,
      "step": 6423
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.489565134048462,
      "learning_rate": 3.2094075228160656e-08,
      "loss": 0.402,
      "step": 6424
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.393269419670105,
      "learning_rate": 3.1819472714686994e-08,
      "loss": 0.4386,
      "step": 6425
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1260722875595093,
      "learning_rate": 3.1546046277605716e-08,
      "loss": 0.3896,
      "step": 6426
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1493836641311646,
      "learning_rate": 3.127379598164282e-08,
      "loss": 0.4166,
      "step": 6427
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4026379585266113,
      "learning_rate": 3.100272189124509e-08,
      "loss": 0.4052,
      "step": 6428
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1427381038665771,
      "learning_rate": 3.0732824070582314e-08,
      "loss": 0.4151,
      "step": 6429
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0112407207489014,
      "learning_rate": 3.0464102583545595e-08,
      "loss": 0.4366,
      "step": 6430
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4084118604660034,
      "learning_rate": 3.0196557493745726e-08,
      "loss": 0.4263,
      "step": 6431
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5577572584152222,
      "learning_rate": 2.9930188864518153e-08,
      "loss": 0.4787,
      "step": 6432
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3754900693893433,
      "learning_rate": 2.9664996758916874e-08,
      "loss": 0.421,
      "step": 6433
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2656077146530151,
      "learning_rate": 2.9400981239719463e-08,
      "loss": 0.4922,
      "step": 6434
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5121854543685913,
      "learning_rate": 2.9138142369423694e-08,
      "loss": 0.4641,
      "step": 6435
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3966081142425537,
      "learning_rate": 2.8876480210248693e-08,
      "loss": 0.489,
      "step": 6436
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5749725103378296,
      "learning_rate": 2.8615994824136572e-08,
      "loss": 0.3992,
      "step": 6437
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.106361746788025,
      "learning_rate": 2.8356686272749678e-08,
      "loss": 0.4869,
      "step": 6438
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0558873414993286,
      "learning_rate": 2.8098554617471685e-08,
      "loss": 0.4131,
      "step": 6439
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2813559770584106,
      "learning_rate": 2.78415999194076e-08,
      "loss": 0.3809,
      "step": 6440
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2929741144180298,
      "learning_rate": 2.7585822239384884e-08,
      "loss": 0.4652,
      "step": 6441
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4342318773269653,
      "learning_rate": 2.7331221637951767e-08,
      "loss": 0.4455,
      "step": 6442
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.0545952320098877,
      "learning_rate": 2.7077798175377256e-08,
      "loss": 0.454,
      "step": 6443
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.728029727935791,
      "learning_rate": 2.6825551911652813e-08,
      "loss": 0.4338,
      "step": 6444
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.33897066116333,
      "learning_rate": 2.6574482906489563e-08,
      "loss": 0.3664,
      "step": 6445
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.556009292602539,
      "learning_rate": 2.6324591219322183e-08,
      "loss": 0.5091,
      "step": 6446
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.378282904624939,
      "learning_rate": 2.6075876909304465e-08,
      "loss": 0.5622,
      "step": 6447
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.332837700843811,
      "learning_rate": 2.5828340035313205e-08,
      "loss": 0.4111,
      "step": 6448
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.213096261024475,
      "learning_rate": 2.5581980655945415e-08,
      "loss": 0.4032,
      "step": 6449
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4635943174362183,
      "learning_rate": 2.5336798829519448e-08,
      "loss": 0.4278,
      "step": 6450
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1515957117080688,
      "learning_rate": 2.50927946140761e-08,
      "loss": 0.4256,
      "step": 6451
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.322031855583191,
      "learning_rate": 2.484996806737583e-08,
      "loss": 0.4985,
      "step": 6452
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.110062599182129,
      "learning_rate": 2.4608319246900436e-08,
      "loss": 0.6852,
      "step": 6453
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4732545614242554,
      "learning_rate": 2.4367848209854717e-08,
      "loss": 0.4455,
      "step": 6454
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5507073402404785,
      "learning_rate": 2.412855501316258e-08,
      "loss": 0.4695,
      "step": 6455
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3559330701828003,
      "learning_rate": 2.3890439713470382e-08,
      "loss": 0.4802,
      "step": 6456
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.065377116203308,
      "learning_rate": 2.36535023671447e-08,
      "loss": 0.6763,
      "step": 6457
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.941843867301941,
      "learning_rate": 2.3417743030274e-08,
      "loss": 0.4446,
      "step": 6458
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2880966663360596,
      "learning_rate": 2.318316175866697e-08,
      "loss": 0.4476,
      "step": 6459
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0779924392700195,
      "learning_rate": 2.294975860785531e-08,
      "loss": 0.4285,
      "step": 6460
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0479294061660767,
      "learning_rate": 2.2717533633090373e-08,
      "loss": 0.4276,
      "step": 6461
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2236602306365967,
      "learning_rate": 2.2486486889343738e-08,
      "loss": 0.4506,
      "step": 6462
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1901326179504395,
      "learning_rate": 2.225661843131055e-08,
      "loss": 0.4556,
      "step": 6463
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3176002502441406,
      "learning_rate": 2.2027928313405056e-08,
      "loss": 0.4675,
      "step": 6464
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4057717323303223,
      "learning_rate": 2.180041658976284e-08,
      "loss": 0.4285,
      "step": 6465
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6726332902908325,
      "learning_rate": 2.1574083314241933e-08,
      "loss": 0.5032,
      "step": 6466
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2303828001022339,
      "learning_rate": 2.1348928540418923e-08,
      "loss": 0.4333,
      "step": 6467
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0585397481918335,
      "learning_rate": 2.1124952321594505e-08,
      "loss": 0.4615,
      "step": 6468
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8582974672317505,
      "learning_rate": 2.090215471078738e-08,
      "loss": 0.4866,
      "step": 6469
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.1697046756744385,
      "learning_rate": 2.0680535760738695e-08,
      "loss": 0.4274,
      "step": 6470
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4058775901794434,
      "learning_rate": 2.0460095523910928e-08,
      "loss": 0.4496,
      "step": 6471
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.217549443244934,
      "learning_rate": 2.0240834052487335e-08,
      "loss": 0.4871,
      "step": 6472
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6864676475524902,
      "learning_rate": 2.0022751398371397e-08,
      "loss": 0.47,
      "step": 6473
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2458324432373047,
      "learning_rate": 1.9805847613187936e-08,
      "loss": 0.4886,
      "step": 6474
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.396544337272644,
      "learning_rate": 1.959012274828309e-08,
      "loss": 0.41,
      "step": 6475
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3137986660003662,
      "learning_rate": 1.93755768547238e-08,
      "loss": 0.5223,
      "step": 6476
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3749853372573853,
      "learning_rate": 1.916220998329721e-08,
      "loss": 0.4765,
      "step": 6477
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2159457206726074,
      "learning_rate": 1.895002218451236e-08,
      "loss": 0.471,
      "step": 6478
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2949837446212769,
      "learning_rate": 1.873901350859797e-08,
      "loss": 0.4405,
      "step": 6479
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.627358078956604,
      "learning_rate": 1.8529184005505184e-08,
      "loss": 0.4196,
      "step": 6480
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0214987993240356,
      "learning_rate": 1.832053372490483e-08,
      "loss": 0.4814,
      "step": 6481
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.018399715423584,
      "learning_rate": 1.8113062716189622e-08,
      "loss": 0.4111,
      "step": 6482
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1102806329727173,
      "learning_rate": 1.790677102847138e-08,
      "loss": 0.4654,
      "step": 6483
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.553207516670227,
      "learning_rate": 1.7701658710584935e-08,
      "loss": 0.3844,
      "step": 6484
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.262701153755188,
      "learning_rate": 1.7497725811083665e-08,
      "loss": 0.4314,
      "step": 6485
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2130650281906128,
      "learning_rate": 1.7294972378243957e-08,
      "loss": 0.7118,
      "step": 6486
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.264988660812378,
      "learning_rate": 1.7093398460061307e-08,
      "loss": 0.505,
      "step": 6487
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5971143245697021,
      "learning_rate": 1.6893004104253095e-08,
      "loss": 0.501,
      "step": 6488
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9120880365371704,
      "learning_rate": 1.669378935825694e-08,
      "loss": 0.4509,
      "step": 6489
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4937236309051514,
      "learning_rate": 1.6495754269230668e-08,
      "loss": 0.4122,
      "step": 6490
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9696075916290283,
      "learning_rate": 1.629889888405456e-08,
      "loss": 0.5109,
      "step": 6491
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2614604234695435,
      "learning_rate": 1.6103223249327447e-08,
      "loss": 0.4403,
      "step": 6492
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0947248935699463,
      "learning_rate": 1.5908727411371172e-08,
      "loss": 0.6973,
      "step": 6493
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3183964490890503,
      "learning_rate": 1.571541141622612e-08,
      "loss": 0.4259,
      "step": 6494
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.20807683467865,
      "learning_rate": 1.5523275309654584e-08,
      "loss": 0.474,
      "step": 6495
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3740887641906738,
      "learning_rate": 1.5332319137140173e-08,
      "loss": 0.4202,
      "step": 6496
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.1129415035247803,
      "learning_rate": 1.5142542943885618e-08,
      "loss": 0.4619,
      "step": 6497
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0788410902023315,
      "learning_rate": 1.495394677481443e-08,
      "loss": 0.6707,
      "step": 6498
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5587564706802368,
      "learning_rate": 1.4766530674573122e-08,
      "loss": 0.5589,
      "step": 6499
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2228120565414429,
      "learning_rate": 1.4580294687526198e-08,
      "loss": 0.4086,
      "step": 6500
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.858881950378418,
      "learning_rate": 1.4395238857759508e-08,
      "loss": 0.4524,
      "step": 6501
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5836122035980225,
      "learning_rate": 1.4211363229080233e-08,
      "loss": 0.4434,
      "step": 6502
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.9333999156951904,
      "learning_rate": 1.4028667845015775e-08,
      "loss": 0.4137,
      "step": 6503
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2748554944992065,
      "learning_rate": 1.3847152748813763e-08,
      "loss": 0.4175,
      "step": 6504
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.351900577545166,
      "learning_rate": 1.3666817983442604e-08,
      "loss": 0.4343,
      "step": 6505
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.279403805732727,
      "learning_rate": 1.3487663591592037e-08,
      "loss": 0.417,
      "step": 6506
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2113673686981201,
      "learning_rate": 1.3309689615672029e-08,
      "loss": 0.4558,
      "step": 6507
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5240521430969238,
      "learning_rate": 1.3132896097811653e-08,
      "loss": 0.5021,
      "step": 6508
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5810916423797607,
      "learning_rate": 1.2957283079862992e-08,
      "loss": 0.5109,
      "step": 6509
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.069261074066162,
      "learning_rate": 1.2782850603397234e-08,
      "loss": 0.4837,
      "step": 6510
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3368276357650757,
      "learning_rate": 1.2609598709705794e-08,
      "loss": 0.5101,
      "step": 6511
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8621834516525269,
      "learning_rate": 1.2437527439801422e-08,
      "loss": 0.4576,
      "step": 6512
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1470112800598145,
      "learning_rate": 1.2266636834417645e-08,
      "loss": 0.4678,
      "step": 6513
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.309768795967102,
      "learning_rate": 1.2096926934007103e-08,
      "loss": 0.523,
      "step": 6514
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0391721725463867,
      "learning_rate": 1.1928397778744327e-08,
      "loss": 0.4399,
      "step": 6515
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2172378301620483,
      "learning_rate": 1.1761049408524072e-08,
      "loss": 0.6826,
      "step": 6516
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2865697145462036,
      "learning_rate": 1.159488186296076e-08,
      "loss": 0.4926,
      "step": 6517
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1900835037231445,
      "learning_rate": 1.1429895181390704e-08,
      "loss": 0.7098,
      "step": 6518
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2565420866012573,
      "learning_rate": 1.1266089402868773e-08,
      "loss": 0.5005,
      "step": 6519
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.359850287437439,
      "learning_rate": 1.1103464566171729e-08,
      "loss": 0.4643,
      "step": 6520
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2808176279067993,
      "learning_rate": 1.0942020709797107e-08,
      "loss": 0.44,
      "step": 6521
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.0109100341796875,
      "learning_rate": 1.0781757871961007e-08,
      "loss": 0.4213,
      "step": 6522
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5977429151535034,
      "learning_rate": 1.0622676090601969e-08,
      "loss": 0.4847,
      "step": 6523
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.059293508529663,
      "learning_rate": 1.0464775403378202e-08,
      "loss": 0.3438,
      "step": 6524
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1316790580749512,
      "learning_rate": 1.0308055847667586e-08,
      "loss": 0.423,
      "step": 6525
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4809373617172241,
      "learning_rate": 1.0152517460569333e-08,
      "loss": 0.4187,
      "step": 6526
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1359832286834717,
      "learning_rate": 9.99816027890288e-09,
      "loss": 0.3747,
      "step": 6527
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5019869804382324,
      "learning_rate": 9.844984339207886e-09,
      "loss": 0.4371,
      "step": 6528
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.821599245071411,
      "learning_rate": 9.692989677744235e-09,
      "loss": 0.3618,
      "step": 6529
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.449866533279419,
      "learning_rate": 9.542176330493147e-09,
      "loss": 0.4754,
      "step": 6530
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3230857849121094,
      "learning_rate": 9.392544333154397e-09,
      "loss": 0.4616,
      "step": 6531
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4619239568710327,
      "learning_rate": 9.244093721149094e-09,
      "loss": 0.4118,
      "step": 6532
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3604918718338013,
      "learning_rate": 9.096824529619686e-09,
      "loss": 0.4192,
      "step": 6533
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.109183669090271,
      "learning_rate": 8.950736793427172e-09,
      "loss": 0.4148,
      "step": 6534
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6537081003189087,
      "learning_rate": 8.805830547154448e-09,
      "loss": 0.4553,
      "step": 6535
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2682186365127563,
      "learning_rate": 8.662105825103518e-09,
      "loss": 0.567,
      "step": 6536
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2023403644561768,
      "learning_rate": 8.51956266129661e-09,
      "loss": 0.506,
      "step": 6537
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3431918621063232,
      "learning_rate": 8.378201089477844e-09,
      "loss": 0.576,
      "step": 6538
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5372483730316162,
      "learning_rate": 8.238021143109898e-09,
      "loss": 0.4694,
      "step": 6539
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.039552688598633,
      "learning_rate": 8.099022855376782e-09,
      "loss": 0.5027,
      "step": 6540
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2795366048812866,
      "learning_rate": 7.96120625918162e-09,
      "loss": 0.461,
      "step": 6541
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1053409576416016,
      "learning_rate": 7.824571387149982e-09,
      "loss": 0.4762,
      "step": 6542
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1958609819412231,
      "learning_rate": 7.68911827162544e-09,
      "loss": 0.5241,
      "step": 6543
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.309142827987671,
      "learning_rate": 7.554846944672345e-09,
      "loss": 0.4838,
      "step": 6544
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6177839040756226,
      "learning_rate": 7.421757438076382e-09,
      "loss": 0.4227,
      "step": 6545
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4052574634552002,
      "learning_rate": 7.289849783342351e-09,
      "loss": 0.487,
      "step": 6546
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.111006498336792,
      "learning_rate": 7.159124011695828e-09,
      "loss": 0.4832,
      "step": 6547
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.233916997909546,
      "learning_rate": 7.029580154082061e-09,
      "loss": 0.4221,
      "step": 6548
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2023818492889404,
      "learning_rate": 6.901218241167629e-09,
      "loss": 0.4221,
      "step": 6549
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.583094835281372,
      "learning_rate": 6.774038303338226e-09,
      "loss": 0.4547,
      "step": 6550
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2069650888442993,
      "learning_rate": 6.648040370699771e-09,
      "loss": 0.427,
      "step": 6551
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0586822032928467,
      "learning_rate": 6.52322447307896e-09,
      "loss": 0.4443,
      "step": 6552
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4708023071289062,
      "learning_rate": 6.3995906400221575e-09,
      "loss": 0.45,
      "step": 6553
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.669519305229187,
      "learning_rate": 6.277138900797064e-09,
      "loss": 0.4318,
      "step": 6554
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5437222719192505,
      "learning_rate": 6.155869284389937e-09,
      "loss": 0.4332,
      "step": 6555
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1510725021362305,
      "learning_rate": 6.035781819507813e-09,
      "loss": 0.4028,
      "step": 6556
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4900295734405518,
      "learning_rate": 5.916876534578508e-09,
      "loss": 0.467,
      "step": 6557
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2118873596191406,
      "learning_rate": 5.7991534577495065e-09,
      "loss": 0.4591,
      "step": 6558
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.512310028076172,
      "learning_rate": 5.682612616887961e-09,
      "loss": 0.3951,
      "step": 6559
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3294638395309448,
      "learning_rate": 5.567254039581804e-09,
      "loss": 0.4074,
      "step": 6560
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.159542202949524,
      "learning_rate": 5.453077753139191e-09,
      "loss": 0.4574,
      "step": 6561
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.588291883468628,
      "learning_rate": 5.3400837845873906e-09,
      "loss": 0.4279,
      "step": 6562
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6073310375213623,
      "learning_rate": 5.228272160675563e-09,
      "loss": 0.4457,
      "step": 6563
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0081006288528442,
      "learning_rate": 5.117642907870868e-09,
      "loss": 0.4173,
      "step": 6564
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3533374071121216,
      "learning_rate": 5.008196052362912e-09,
      "loss": 0.4899,
      "step": 6565
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2754145860671997,
      "learning_rate": 4.899931620058751e-09,
      "loss": 0.3946,
      "step": 6566
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6666414737701416,
      "learning_rate": 4.792849636588437e-09,
      "loss": 0.4836,
      "step": 6567
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.141700029373169,
      "learning_rate": 4.686950127298917e-09,
      "loss": 0.4308,
      "step": 6568
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5024633407592773,
      "learning_rate": 4.582233117260693e-09,
      "loss": 0.4573,
      "step": 6569
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2588709592819214,
      "learning_rate": 4.478698631260603e-09,
      "loss": 0.5139,
      "step": 6570
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5890792608261108,
      "learning_rate": 4.3763466938095965e-09,
      "loss": 0.4808,
      "step": 6571
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1888508796691895,
      "learning_rate": 4.275177329134961e-09,
      "loss": 0.4762,
      "step": 6572
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8273193836212158,
      "learning_rate": 4.175190561186426e-09,
      "loss": 0.5642,
      "step": 6573
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2530474662780762,
      "learning_rate": 4.076386413632838e-09,
      "loss": 0.4324,
      "step": 6574
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.5393567085266113,
      "learning_rate": 3.978764909863264e-09,
      "loss": 0.482,
      "step": 6575
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3441611528396606,
      "learning_rate": 3.882326072987552e-09,
      "loss": 0.5081,
      "step": 6576
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.100043535232544,
      "learning_rate": 3.787069925833553e-09,
      "loss": 0.444,
      "step": 6577
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.5685932636260986,
      "learning_rate": 3.6929964909515615e-09,
      "loss": 0.4956,
      "step": 6578
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4145095348358154,
      "learning_rate": 3.6001057906104307e-09,
      "loss": 0.4966,
      "step": 6579
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.6814377307891846,
      "learning_rate": 3.5083978467992384e-09,
      "loss": 0.4703,
      "step": 6580
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6719717979431152,
      "learning_rate": 3.4178726812272857e-09,
      "loss": 0.4877,
      "step": 6581
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.333673357963562,
      "learning_rate": 3.3285303153240965e-09,
      "loss": 0.4412,
      "step": 6582
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6336694955825806,
      "learning_rate": 3.2403707702388653e-09,
      "loss": 0.4503,
      "step": 6583
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2184507846832275,
      "learning_rate": 3.153394066840454e-09,
      "loss": 0.4712,
      "step": 6584
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2235267162322998,
      "learning_rate": 3.0676002257190583e-09,
      "loss": 0.4456,
      "step": 6585
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.222869873046875,
      "learning_rate": 2.9829892671828785e-09,
      "loss": 0.488,
      "step": 6586
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2226697206497192,
      "learning_rate": 2.899561211262003e-09,
      "loss": 0.4122,
      "step": 6587
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4674283266067505,
      "learning_rate": 2.817316077705079e-09,
      "loss": 0.447,
      "step": 6588
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4075546264648438,
      "learning_rate": 2.7362538859820873e-09,
      "loss": 0.5073,
      "step": 6589
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5489782094955444,
      "learning_rate": 2.6563746552815683e-09,
      "loss": 0.4478,
      "step": 6590
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1331112384796143,
      "learning_rate": 2.5776784045128402e-09,
      "loss": 0.4671,
      "step": 6591
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7166048288345337,
      "learning_rate": 2.5001651523048897e-09,
      "loss": 0.4531,
      "step": 6592
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.156693458557129,
      "learning_rate": 2.4238349170069287e-09,
      "loss": 0.3935,
      "step": 6593
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5183659791946411,
      "learning_rate": 2.3486877166878363e-09,
      "loss": 0.4915,
      "step": 6594
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.195923924446106,
      "learning_rate": 2.2747235691367164e-09,
      "loss": 0.4178,
      "step": 6595
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9703136682510376,
      "learning_rate": 2.201942491862341e-09,
      "loss": 0.6581,
      "step": 6596
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.283877968788147,
      "learning_rate": 2.1303445020937063e-09,
      "loss": 0.7422,
      "step": 6597
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2648640871047974,
      "learning_rate": 2.059929616780032e-09,
      "loss": 0.357,
      "step": 6598
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2246431112289429,
      "learning_rate": 1.9906978525896513e-09,
      "loss": 0.475,
      "step": 6599
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1181622743606567,
      "learning_rate": 1.9226492259111216e-09,
      "loss": 0.422,
      "step": 6600
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2608575820922852,
      "learning_rate": 1.855783752852669e-09,
      "loss": 0.4529,
      "step": 6601
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2589470148086548,
      "learning_rate": 1.7901014492438528e-09,
      "loss": 0.5123,
      "step": 6602
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2315824031829834,
      "learning_rate": 1.7256023306322366e-09,
      "loss": 0.4717,
      "step": 6603
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.561318278312683,
      "learning_rate": 1.6622864122867176e-09,
      "loss": 0.5065,
      "step": 6604
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.228587031364441,
      "learning_rate": 1.6001537091947518e-09,
      "loss": 0.4121,
      "step": 6605
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2897117137908936,
      "learning_rate": 1.5392042360656834e-09,
      "loss": 0.4776,
      "step": 6606
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.327803373336792,
      "learning_rate": 1.479438007326306e-09,
      "loss": 0.5054,
      "step": 6607
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.057539939880371,
      "learning_rate": 1.4208550371258568e-09,
      "loss": 0.696,
      "step": 6608
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4263715744018555,
      "learning_rate": 1.3634553393310213e-09,
      "loss": 0.4528,
      "step": 6609
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2088463306427002,
      "learning_rate": 1.3072389275303742e-09,
      "loss": 0.7136,
      "step": 6610
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4601577520370483,
      "learning_rate": 1.252205815031049e-09,
      "loss": 0.4806,
      "step": 6611
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2029852867126465,
      "learning_rate": 1.1983560148609575e-09,
      "loss": 0.7295,
      "step": 6612
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4111748933792114,
      "learning_rate": 1.1456895397671253e-09,
      "loss": 0.5293,
      "step": 6613
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2572365999221802,
      "learning_rate": 1.094206402217357e-09,
      "loss": 0.4896,
      "step": 6614
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1673884391784668,
      "learning_rate": 1.043906614398571e-09,
      "loss": 0.4834,
      "step": 6615
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.73297119140625,
      "learning_rate": 9.94790188217909e-10,
      "loss": 0.4643,
      "step": 6616
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1973410844802856,
      "learning_rate": 9.468571353021816e-10,
      "loss": 0.4636,
      "step": 6617
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.898768424987793,
      "learning_rate": 9.001074669978682e-10,
      "loss": 0.3854,
      "step": 6618
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.989920973777771,
      "learning_rate": 8.545411943722271e-10,
      "loss": 0.4056,
      "step": 6619
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.186369776725769,
      "learning_rate": 8.101583282110747e-10,
      "loss": 0.4188,
      "step": 6620
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3609719276428223,
      "learning_rate": 7.669588790221172e-10,
      "loss": 0.4345,
      "step": 6621
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6582825183868408,
      "learning_rate": 7.249428570299532e-10,
      "loss": 0.4745,
      "step": 6622
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.483932614326477,
      "learning_rate": 6.841102721821813e-10,
      "loss": 0.5351,
      "step": 6623
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5367463827133179,
      "learning_rate": 6.444611341432927e-10,
      "loss": 0.5314,
      "step": 6624
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.119049072265625,
      "learning_rate": 6.059954523002232e-10,
      "loss": 0.4346,
      "step": 6625
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5325359106063843,
      "learning_rate": 5.687132357584669e-10,
      "loss": 0.4873,
      "step": 6626
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6526154279708862,
      "learning_rate": 5.326144933431864e-10,
      "loss": 0.4308,
      "step": 6627
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5596330165863037,
      "learning_rate": 4.976992336003239e-10,
      "loss": 0.4205,
      "step": 6628
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.2380640506744385,
      "learning_rate": 4.6396746479382413e-10,
      "loss": 0.5101,
      "step": 6629
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.60537588596344,
      "learning_rate": 4.3141919491007655e-10,
      "loss": 0.5139,
      "step": 6630
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3016064167022705,
      "learning_rate": 4.000544316534738e-10,
      "loss": 0.3813,
      "step": 6631
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5514304637908936,
      "learning_rate": 3.6987318244863234e-10,
      "loss": 0.3471,
      "step": 6632
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2404531240463257,
      "learning_rate": 3.408754544403925e-10,
      "loss": 0.4471,
      "step": 6633
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4376779794692993,
      "learning_rate": 3.130612544927081e-10,
      "loss": 0.4192,
      "step": 6634
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2511579990386963,
      "learning_rate": 2.864305891903119e-10,
      "loss": 0.5314,
      "step": 6635
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0797035694122314,
      "learning_rate": 2.609834648364951e-10,
      "loss": 0.5067,
      "step": 6636
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2793300151824951,
      "learning_rate": 2.3671988745588294e-10,
      "loss": 0.448,
      "step": 6637
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4702308177947998,
      "learning_rate": 2.1363986279221428e-10,
      "loss": 0.4915,
      "step": 6638
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2369475364685059,
      "learning_rate": 1.917433963083415e-10,
      "loss": 0.455,
      "step": 6639
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4297107458114624,
      "learning_rate": 1.71030493188451e-10,
      "loss": 0.4133,
      "step": 6640
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4607726335525513,
      "learning_rate": 1.5150115833528766e-10,
      "loss": 0.414,
      "step": 6641
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3912570476531982,
      "learning_rate": 1.3315539637182018e-10,
      "loss": 0.4963,
      "step": 6642
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.356311321258545,
      "learning_rate": 1.15993211641241e-10,
      "loss": 0.4806,
      "step": 6643
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7494667768478394,
      "learning_rate": 1.0001460820530107e-10,
      "loss": 0.4975,
      "step": 6644
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3961743116378784,
      "learning_rate": 8.521958984819556e-11,
      "loss": 0.428,
      "step": 6645
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1013633012771606,
      "learning_rate": 7.160816007045768e-11,
      "loss": 0.4977,
      "step": 6646
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3433891534805298,
      "learning_rate": 5.918032209506486e-11,
      "loss": 0.4848,
      "step": 6647
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.116370677947998,
      "learning_rate": 4.7936078864108116e-11,
      "loss": 0.4442,
      "step": 6648
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4249285459518433,
      "learning_rate": 3.787543303879204e-11,
      "loss": 0.452,
      "step": 6649
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.5555520057678223,
      "learning_rate": 2.8998387001100137e-11,
      "loss": 0.4442,
      "step": 6650
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.8289107084274292,
      "learning_rate": 2.1304942852129472e-11,
      "loss": 0.4353,
      "step": 6651
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.1057546138763428,
      "learning_rate": 1.4795102413200924e-11,
      "loss": 0.4097,
      "step": 6652
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.057705044746399,
      "learning_rate": 9.468867225859158e-12,
      "loss": 0.5063,
      "step": 6653
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4950459003448486,
      "learning_rate": 5.3262385496521965e-12,
      "loss": 0.5044,
      "step": 6654
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1403788328170776,
      "learning_rate": 2.3672173665723052e-12,
      "loss": 0.4462,
      "step": 6655
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4435445070266724,
      "learning_rate": 5.918043766151016e-13,
      "loss": 0.4247,
      "step": 6656
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.241782546043396,
      "learning_rate": 0.0,
      "loss": 0.6316,
      "step": 6657
    },
    {
      "epoch": 1.0,
      "step": 6657,
      "total_flos": 9.734557311136956e+18,
      "train_loss": 0.4873377132231326,
      "train_runtime": 20189.146,
      "train_samples_per_second": 42.206,
      "train_steps_per_second": 0.33
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 6657,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "total_flos": 9.734557311136956e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}