{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.3333333333333333,
  "eval_steps": 500,
  "global_step": 8100,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0002,
      "loss": 1.8171,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.0703125,
      "learning_rate": 0.0002,
      "loss": 1.815,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0002,
      "loss": 1.8213,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 1.9109,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.921,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.5387,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.7638,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 1.7497,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.8964,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0002,
      "loss": 1.7395,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0002,
      "loss": 1.5338,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002,
      "loss": 1.5693,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.6814,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.2875,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.6479,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.6474,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.3501,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.6771,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.1564,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.5622,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.8172,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3359,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.253,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.2689,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 1.1975,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0002,
      "loss": 0.8924,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.0497,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.3952,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.3515,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.3258,
      "step": 30
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.4702,
      "step": 31
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.7788,
      "step": 32
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.8258,
      "step": 33
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.0249,
      "step": 34
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.1205,
      "step": 35
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.7237,
      "step": 36
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.3971,
      "step": 37
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.1552,
      "step": 38
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.2618,
      "step": 39
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.6461,
      "step": 40
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.2039,
      "step": 41
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.0456,
      "step": 42
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.1877,
      "step": 43
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.355,
      "step": 44
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.3928,
      "step": 45
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.5112,
      "step": 46
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2138,
      "step": 47
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.293,
      "step": 48
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.0777,
      "step": 49
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.5465,
      "step": 50
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 0.9537,
      "step": 51
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 1.1702,
      "step": 52
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.3573,
      "step": 53
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 0.7498,
      "step": 54
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.2369,
      "step": 55
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0002,
      "loss": 1.5412,
      "step": 56
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.365,
      "step": 57
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.2201,
      "step": 58
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002,
      "loss": 0.7067,
      "step": 59
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.5673,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.1418,
      "step": 61
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.0338,
      "step": 62
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.9413,
      "step": 63
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.1271,
      "step": 64
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.3293,
      "step": 65
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3756,
      "step": 66
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 1.104,
      "step": 67
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.0217,
      "step": 68
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.3385,
      "step": 69
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.4918,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.3872,
      "step": 71
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.2167,
      "step": 72
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.0748,
      "step": 73
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.2968,
      "step": 74
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.1125,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.0673,
      "step": 76
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.4125,
      "step": 77
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.2796,
      "step": 78
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 1.3073,
      "step": 79
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.1468,
      "step": 80
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 1.1405,
      "step": 81
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.2268,
      "step": 82
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 1.0195,
      "step": 83
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.9584,
      "step": 84
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2071,
      "step": 85
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.4299,
      "step": 86
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.5783,
      "step": 87
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.8598,
      "step": 88
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.2615,
      "step": 89
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.8759,
      "step": 90
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 1.3278,
      "step": 91
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.1526,
      "step": 92
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.2039,
      "step": 93
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.1265,
      "step": 94
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.3839,
      "step": 95
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.4274,
      "step": 96
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.1425,
      "step": 97
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1982,
      "step": 98
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.0613,
      "step": 99
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.2323,
      "step": 100
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.5413,
      "step": 101
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.6045,
      "step": 102
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2584,
      "step": 103
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.8248,
      "step": 104
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.2307,
      "step": 105
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.0323,
      "step": 106
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.163,
      "step": 107
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.448,
      "step": 108
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0002,
      "loss": 1.103,
      "step": 109
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.3462,
      "step": 110
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.7998,
      "step": 111
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.0399,
      "step": 112
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 0.9893,
      "step": 113
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.358,
      "step": 114
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.418,
      "step": 115
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.9144,
      "step": 116
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.4713,
      "step": 117
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.1704,
      "step": 118
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.1326,
      "step": 119
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.3675,
      "step": 120
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 1.2578,
      "step": 121
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.2701,
      "step": 122
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.9934,
      "step": 123
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 0.9657,
      "step": 124
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002,
      "loss": 0.5736,
      "step": 125
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.4506,
      "step": 126
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.1503,
      "step": 127
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.1589,
      "step": 128
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.8476,
      "step": 129
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.4279,
      "step": 130
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.8599,
      "step": 131
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.8042,
      "step": 132
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.2247,
      "step": 133
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.6425,
      "step": 134
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 0.917,
      "step": 135
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.0768,
      "step": 136
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.5944,
      "step": 137
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3733,
      "step": 138
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0002,
      "loss": 0.5677,
      "step": 139
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.4019,
      "step": 140
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 0.9652,
      "step": 141
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.0586,
      "step": 142
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.1658,
      "step": 143
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.3105,
      "step": 144
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.204,
      "step": 145
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 0.9472,
      "step": 146
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 0.9385,
      "step": 147
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.256,
      "step": 148
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.4043,
      "step": 149
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0002,
      "loss": 1.3866,
      "step": 150
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.1163,
      "step": 151
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.282,
      "step": 152
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.1188,
      "step": 153
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.0974,
      "step": 154
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.2984,
      "step": 155
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.2658,
      "step": 156
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.2573,
      "step": 157
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.105,
      "step": 158
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0464,
      "step": 159
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.2602,
      "step": 160
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.2196,
      "step": 161
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.3243,
      "step": 162
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.0243,
      "step": 163
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.001,
      "step": 164
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.9044,
      "step": 165
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0497,
      "step": 166
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.3777,
      "step": 167
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.9473,
      "step": 168
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.2238,
      "step": 169
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.221,
      "step": 170
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.4139,
      "step": 171
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.4757,
      "step": 172
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.9818,
      "step": 173
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.2409,
      "step": 174
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 1.0688,
      "step": 175
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.5211,
      "step": 176
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.4223,
      "step": 177
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.16,
      "step": 178
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.9692,
      "step": 179
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.314,
      "step": 180
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.2584,
      "step": 181
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.0019,
      "step": 182
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.0845,
      "step": 183
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.2511,
      "step": 184
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.0775,
      "step": 185
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1525,
      "step": 186
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.9412,
      "step": 187
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.214,
      "step": 188
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.395,
      "step": 189
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.2548,
      "step": 190
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.6418,
      "step": 191
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.0408,
      "step": 192
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 0.9659,
      "step": 193
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 0.9032,
      "step": 194
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.1754,
      "step": 195
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1652,
      "step": 196
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.3417,
      "step": 197
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.2111,
      "step": 198
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.3402,
      "step": 199
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1613,
      "step": 200
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.3982,
      "step": 201
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1378,
      "step": 202
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 0.8756,
      "step": 203
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.1283,
      "step": 204
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.3062,
      "step": 205
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.3789,
      "step": 206
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 0.8405,
      "step": 207
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.6054,
      "step": 208
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.4206,
      "step": 209
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.8477,
      "step": 210
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.0629,
      "step": 211
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.3358,
      "step": 212
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 0.9821,
      "step": 213
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.0477,
      "step": 214
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.8535,
      "step": 215
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.092,
      "step": 216
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.0195,
      "step": 217
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.1347,
      "step": 218
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.3226,
      "step": 219
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 0.9919,
      "step": 220
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.1372,
      "step": 221
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 0.9801,
      "step": 222
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.8933,
      "step": 223
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.4091,
      "step": 224
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.5164,
      "step": 225
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.0949,
      "step": 226
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.588,
      "step": 227
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.1051,
      "step": 228
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.0262,
      "step": 229
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.1759,
      "step": 230
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.0477,
      "step": 231
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.0756,
      "step": 232
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.4234,
      "step": 233
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.177,
      "step": 234
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.0279,
      "step": 235
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.2868,
      "step": 236
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.119,
      "step": 237
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.3527,
      "step": 238
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 1.2913,
      "step": 239
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.9767,
      "step": 240
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.4048,
      "step": 241
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.3675,
      "step": 242
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.3262,
      "step": 243
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.3715,
      "step": 244
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0002,
      "loss": 1.1929,
      "step": 245
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1472,
      "step": 246
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.9078,
      "step": 247
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.0733,
      "step": 248
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.1299,
      "step": 249
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.4634,
      "step": 250
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 0.6726,
      "step": 251
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.2409,
      "step": 252
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.1537,
      "step": 253
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.0727,
      "step": 254
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 1.1325,
      "step": 255
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.7394,
      "step": 256
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.4477,
      "step": 257
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.0341,
      "step": 258
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.1367,
      "step": 259
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.7778,
      "step": 260
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.3567,
      "step": 261
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.2657,
      "step": 262
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.1944,
      "step": 263
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.0383,
      "step": 264
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.2281,
      "step": 265
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.3164,
      "step": 266
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.0633,
      "step": 267
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0002,
      "loss": 1.2593,
      "step": 268
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.1094,
      "step": 269
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.1848,
      "step": 270
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.1762,
      "step": 271
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.2223,
      "step": 272
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 1.1354,
      "step": 273
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.0866,
      "step": 274
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.6438,
      "step": 275
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.1669,
      "step": 276
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.2979,
      "step": 277
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.2681,
      "step": 278
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.4067,
      "step": 279
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.961,
      "step": 280
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.124,
      "step": 281
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.8956,
      "step": 282
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.0273,
      "step": 283
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.1708,
      "step": 284
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.0383,
      "step": 285
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.0315,
      "step": 286
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.1859,
      "step": 287
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.5163,
      "step": 288
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.9571,
      "step": 289
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.1055,
      "step": 290
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.0859,
      "step": 291
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.2354,
      "step": 292
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.1997,
      "step": 293
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.2524,
      "step": 294
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.9583,
      "step": 295
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.3942,
      "step": 296
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3604,
      "step": 297
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.9933,
      "step": 298
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0897,
      "step": 299
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.3611,
      "step": 300
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.0342,
      "step": 301
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 1.2847,
      "step": 302
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.2567,
      "step": 303
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.0074,
      "step": 304
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.471,
      "step": 305
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.3169,
      "step": 306
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.1973,
      "step": 307
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.5984,
      "step": 308
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.2598,
      "step": 309
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 1.0041,
      "step": 310
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.191,
      "step": 311
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.0422,
      "step": 312
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.1136,
      "step": 313
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.1967,
      "step": 314
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.136,
      "step": 315
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.0245,
      "step": 316
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.5548,
      "step": 317
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.2085,
      "step": 318
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.8588,
      "step": 319
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.3477,
      "step": 320
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 0.9935,
      "step": 321
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0986,
      "step": 322
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.2191,
      "step": 323
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.3792,
      "step": 324
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 0.998,
      "step": 325
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.1994,
      "step": 326
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.0084,
      "step": 327
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.0106,
      "step": 328
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.3684,
      "step": 329
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.4506,
      "step": 330
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.0091,
      "step": 331
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.1408,
      "step": 332
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.9736,
      "step": 333
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.2159,
      "step": 334
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0002,
      "loss": 1.1514,
      "step": 335
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.5979,
      "step": 336
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.1091,
      "step": 337
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.2236,
      "step": 338
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.1744,
      "step": 339
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.949,
      "step": 340
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.1461,
      "step": 341
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.8692,
      "step": 342
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.4037,
      "step": 343
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.2644,
      "step": 344
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.0985,
      "step": 345
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3491,
      "step": 346
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0436,
      "step": 347
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.1722,
      "step": 348
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.995,
      "step": 349
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.0101,
      "step": 350
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.1853,
      "step": 351
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.1055,
      "step": 352
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.956,
      "step": 353
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.0566,
      "step": 354
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.2636,
      "step": 355
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.4228,
      "step": 356
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.4084,
      "step": 357
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.2523,
      "step": 358
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.3957,
      "step": 359
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2305,
      "step": 360
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 1.115,
      "step": 361
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.0908,
      "step": 362
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.4158,
      "step": 363
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 0.6825,
      "step": 364
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.1175,
      "step": 365
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.4711,
      "step": 366
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 1.1005,
      "step": 367
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.4817,
      "step": 368
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.2146,
      "step": 369
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0002,
      "loss": 1.1038,
      "step": 370
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.2647,
      "step": 371
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.6658,
      "step": 372
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3328,
      "step": 373
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.1634,
      "step": 374
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.0412,
      "step": 375
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.3956,
      "step": 376
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0002,
      "loss": 1.1343,
      "step": 377
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.9778,
      "step": 378
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.2387,
      "step": 379
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.1833,
      "step": 380
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.1581,
      "step": 381
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.0977,
      "step": 382
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.4837,
      "step": 383
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.0132,
      "step": 384
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.9757,
      "step": 385
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.4529,
      "step": 386
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.5529,
      "step": 387
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.959,
      "step": 388
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0218,
      "step": 389
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 1.3675,
      "step": 390
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.9301,
      "step": 391
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.3634,
      "step": 392
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 0.937,
      "step": 393
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.7826,
      "step": 394
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 0.9765,
      "step": 395
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 1.2337,
      "step": 396
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 0.9612,
      "step": 397
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.0907,
      "step": 398
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.2492,
      "step": 399
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.4083,
      "step": 400
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 0.9791,
      "step": 401
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2916,
      "step": 402
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2022,
      "step": 403
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002,
      "loss": 1.4074,
      "step": 404
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.4089,
      "step": 405
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.5779,
      "step": 406
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.0181,
      "step": 407
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.9826,
      "step": 408
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 0.939,
      "step": 409
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.3259,
      "step": 410
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.0935,
      "step": 411
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.488,
      "step": 412
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0002,
      "loss": 0.7566,
      "step": 413
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.1166,
      "step": 414
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.0282,
      "step": 415
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.1762,
      "step": 416
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 0.9195,
      "step": 417
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.2271,
      "step": 418
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.4534,
      "step": 419
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.3041,
      "step": 420
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.2174,
      "step": 421
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.4436,
      "step": 422
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.0397,
      "step": 423
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 0.9695,
      "step": 424
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1243,
      "step": 425
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 0.957,
      "step": 426
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.0903,
      "step": 427
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.0317,
      "step": 428
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.0183,
      "step": 429
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 0.8085,
      "step": 430
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.3717,
      "step": 431
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.0474,
      "step": 432
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.3634,
      "step": 433
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.4435,
      "step": 434
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.2835,
      "step": 435
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.3126,
      "step": 436
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.1586,
      "step": 437
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.5915,
      "step": 438
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1688,
      "step": 439
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.0717,
      "step": 440
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.3219,
      "step": 441
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.1379,
      "step": 442
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.0086,
      "step": 443
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 0.9158,
      "step": 444
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.0264,
      "step": 445
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.7237,
      "step": 446
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.3501,
      "step": 447
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.0012,
      "step": 448
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.2172,
      "step": 449
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.955,
      "step": 450
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.181,
      "step": 451
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.2306,
      "step": 452
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.3274,
      "step": 453
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0002,
      "loss": 0.9931,
      "step": 454
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.0341,
      "step": 455
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0002,
      "loss": 0.6167,
      "step": 456
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.9961,
      "step": 457
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.1603,
      "step": 458
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0002,
      "loss": 0.9412,
      "step": 459
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0002,
      "loss": 1.2581,
      "step": 460
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.41,
      "step": 461
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.4618,
      "step": 462
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.0667,
      "step": 463
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.0194,
      "step": 464
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.2725,
      "step": 465
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.0158,
      "step": 466
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.5972,
      "step": 467
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0002,
      "loss": 0.5525,
      "step": 468
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.0805,
      "step": 469
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 0.8351,
      "step": 470
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0683,
      "step": 471
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.9423,
      "step": 472
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 1.2134,
      "step": 473
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.3894,
      "step": 474
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.0347,
      "step": 475
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.9649,
      "step": 476
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.9558,
      "step": 477
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 1.0644,
      "step": 478
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.2181,
      "step": 479
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.4071,
      "step": 480
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 0.8477,
      "step": 481
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.581,
      "step": 482
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.1302,
      "step": 483
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.3325,
      "step": 484
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.0595,
      "step": 485
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.1093,
      "step": 486
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 0.7134,
      "step": 487
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.0979,
      "step": 488
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.3106,
      "step": 489
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0895,
      "step": 490
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.5127,
      "step": 491
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 1.2682,
      "step": 492
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.4127,
      "step": 493
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.1541,
      "step": 494
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.0014,
      "step": 495
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.1229,
      "step": 496
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.0573,
      "step": 497
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 0.9514,
      "step": 498
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 0.9298,
      "step": 499
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.3909,
      "step": 500
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.037,
      "step": 501
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.6835,
      "step": 502
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.6255,
      "step": 503
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 1.3978,
      "step": 504
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.985,
      "step": 505
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.0948,
      "step": 506
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.3112,
      "step": 507
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.2118,
      "step": 508
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.7211,
      "step": 509
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.2815,
      "step": 510
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.3462,
      "step": 511
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.34,
      "step": 512
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.5453,
      "step": 513
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.2525,
      "step": 514
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.1318,
      "step": 515
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.0462,
      "step": 516
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 1.3552,
      "step": 517
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.2018,
      "step": 518
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.3123,
      "step": 519
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.4213,
      "step": 520
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0002,
      "loss": 0.9244,
      "step": 521
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.5978,
      "step": 522
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.352,
      "step": 523
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 0.6433,
      "step": 524
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.2027,
      "step": 525
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 0.9722,
      "step": 526
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 0.9467,
      "step": 527
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.9993,
      "step": 528
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0002,
      "loss": 0.7283,
      "step": 529
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.1629,
      "step": 530
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.0759,
      "step": 531
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 1.4435,
      "step": 532
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.4303,
      "step": 533
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.9854,
      "step": 534
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0002,
      "loss": 0.977,
      "step": 535
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.3709,
      "step": 536
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.3438,
      "step": 537
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.0399,
      "step": 538
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.153,
      "step": 539
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.0592,
      "step": 540
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 0.8904,
      "step": 541
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 0.8922,
      "step": 542
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.0412,
      "step": 543
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.8543,
      "step": 544
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.9711,
      "step": 545
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.3369,
      "step": 546
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.3074,
      "step": 547
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.1598,
      "step": 548
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.1583,
      "step": 549
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0002,
      "loss": 0.5142,
      "step": 550
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 1.3069,
      "step": 551
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.2606,
      "step": 552
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.2653,
      "step": 553
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.2741,
      "step": 554
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.8914,
      "step": 555
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 0.9402,
      "step": 556
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 1.3362,
      "step": 557
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.387,
      "step": 558
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.128,
      "step": 559
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.2738,
      "step": 560
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2988,
      "step": 561
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2962,
      "step": 562
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 0.973,
      "step": 563
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.2665,
      "step": 564
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.8714,
      "step": 565
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.1623,
      "step": 566
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.3493,
      "step": 567
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.1434,
      "step": 568
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.0027,
      "step": 569
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.2273,
      "step": 570
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.2724,
      "step": 571
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0002,
      "loss": 0.9863,
      "step": 572
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.0394,
      "step": 573
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.4516,
      "step": 574
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.0358,
      "step": 575
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 1.1906,
      "step": 576
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.57,
      "step": 577
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0002,
      "loss": 0.8478,
      "step": 578
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.4664,
      "step": 579
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 1.1292,
      "step": 580
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.2604,
      "step": 581
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 1.0935,
      "step": 582
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 1.2224,
      "step": 583
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.032,
      "step": 584
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.3665,
      "step": 585
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.1331,
      "step": 586
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.2267,
      "step": 587
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 0.9831,
      "step": 588
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.2573,
      "step": 589
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 0.8445,
      "step": 590
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0002,
      "loss": 1.0284,
      "step": 591
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 0.8881,
      "step": 592
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.058,
      "step": 593
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.7563,
      "step": 594
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.2387,
      "step": 595
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.242,
      "step": 596
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.4996,
      "step": 597
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0002,
      "loss": 1.1189,
      "step": 598
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.3419,
      "step": 599
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.3756,
      "step": 600
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.1944,
      "step": 601
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.413,
      "step": 602
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.0139,
      "step": 603
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2527,
      "step": 604
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.0175,
      "step": 605
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.9632,
      "step": 606
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0002,
      "loss": 0.9146,
      "step": 607
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2016,
      "step": 608
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 0.9957,
      "step": 609
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.203,
      "step": 610
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.0709,
      "step": 611
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.6369,
      "step": 612
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.0771,
      "step": 613
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.75,
      "step": 614
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2507,
      "step": 615
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.3101,
      "step": 616
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.9966,
      "step": 617
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.4668,
      "step": 618
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.3248,
      "step": 619
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.7079,
      "step": 620
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.1882,
      "step": 621
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.2078,
      "step": 622
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 0.9609,
      "step": 623
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.3116,
      "step": 624
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.6606,
      "step": 625
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.2722,
      "step": 626
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0002,
      "loss": 0.6843,
      "step": 627
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.0963,
      "step": 628
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0002,
      "loss": 1.1523,
      "step": 629
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.2076,
      "step": 630
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.1309,
      "step": 631
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.4106,
      "step": 632
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.0926,
      "step": 633
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 0.7809,
      "step": 634
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.7264,
      "step": 635
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 0.8426,
      "step": 636
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.2494,
      "step": 637
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 0.8055,
      "step": 638
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.227,
      "step": 639
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0002,
      "loss": 1.178,
      "step": 640
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.2329,
      "step": 641
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.4006,
      "step": 642
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 0.9584,
      "step": 643
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.3137,
      "step": 644
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0002,
      "loss": 1.1797,
      "step": 645
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.9991,
      "step": 646
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.7551,
      "step": 647
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.3593,
      "step": 648
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 0.9464,
      "step": 649
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.3943,
      "step": 650
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.4421,
      "step": 651
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.3466,
      "step": 652
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.6295,
      "step": 653
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.2907,
      "step": 654
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0002,
      "loss": 1.1179,
      "step": 655
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 0.9274,
      "step": 656
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.3988,
      "step": 657
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.0556,
      "step": 658
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.128,
      "step": 659
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.2072,
      "step": 660
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.2676,
      "step": 661
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.8047,
      "step": 662
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.9476,
      "step": 663
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 0.9552,
      "step": 664
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.1108,
      "step": 665
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.4296,
      "step": 666
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.1666,
      "step": 667
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.3502,
      "step": 668
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.3139,
      "step": 669
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.2385,
      "step": 670
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.182,
      "step": 671
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.1803,
      "step": 672
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 0.744,
      "step": 673
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.0538,
      "step": 674
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 0.8663,
      "step": 675
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.1088,
      "step": 676
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.3648,
      "step": 677
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.9221,
      "step": 678
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.8282,
      "step": 679
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.7162,
      "step": 680
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.3016,
      "step": 681
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0002,
      "loss": 1.0631,
      "step": 682
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.0833,
      "step": 683
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0002,
      "loss": 0.9873,
      "step": 684
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.8738,
      "step": 685
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.5323,
      "step": 686
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.0587,
      "step": 687
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.8924,
      "step": 688
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 1.0518,
      "step": 689
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.275,
      "step": 690
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 0.8498,
      "step": 691
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.8331,
      "step": 692
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.2376,
      "step": 693
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.1117,
      "step": 694
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.4263,
      "step": 695
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.0166,
      "step": 696
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.1179,
      "step": 697
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.4038,
      "step": 698
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.4279,
      "step": 699
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.3374,
      "step": 700
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 1.1297,
      "step": 701
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.4205,
      "step": 702
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.5788,
      "step": 703
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.9586,
      "step": 704
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2512,
      "step": 705
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 0.8928,
      "step": 706
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 0.5565,
      "step": 707
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.1688,
      "step": 708
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.287,
      "step": 709
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0002,
      "loss": 0.3227,
      "step": 710
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.3512,
      "step": 711
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 1.1038,
      "step": 712
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.3634,
      "step": 713
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 0.9982,
      "step": 714
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 1.0886,
      "step": 715
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 0.9377,
      "step": 716
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.3115,
      "step": 717
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 0.9763,
      "step": 718
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 0.9932,
      "step": 719
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 0.5658,
      "step": 720
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.7996,
      "step": 721
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0002,
      "loss": 0.9248,
      "step": 722
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.1129,
      "step": 723
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.8857,
      "step": 724
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 0.7768,
      "step": 725
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.339,
      "step": 726
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.0713,
      "step": 727
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.0869,
      "step": 728
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.498,
      "step": 729
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.1878,
      "step": 730
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.8825,
      "step": 731
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.2373,
      "step": 732
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.3826,
      "step": 733
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.8984,
      "step": 734
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.3,
      "step": 735
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.6949,
      "step": 736
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.0645,
      "step": 737
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.2644,
      "step": 738
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0252,
      "step": 739
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.1657,
      "step": 740
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0002,
      "loss": 0.4097,
      "step": 741
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 1.5594,
      "step": 742
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0002,
      "loss": 1.0293,
      "step": 743
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.8279,
      "step": 744
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0002,
      "loss": 1.0735,
      "step": 745
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.2541,
      "step": 746
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.1189,
      "step": 747
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 1.2477,
      "step": 748
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.0563,
      "step": 749
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.4574,
      "step": 750
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.1101,
      "step": 751
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.1468,
      "step": 752
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0002,
      "loss": 0.8638,
      "step": 753
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 1.1233,
      "step": 754
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0002,
      "loss": 1.1963,
      "step": 755
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 0.8509,
      "step": 756
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1138,
      "step": 757
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 0.7904,
      "step": 758
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0874,
      "step": 759
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.2644,
      "step": 760
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.4254,
      "step": 761
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.1682,
      "step": 762
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.9277,
      "step": 763
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.0859375,
      "learning_rate": 0.0002,
      "loss": 0.8483,
      "step": 764
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0002,
      "loss": 1.4765,
      "step": 765
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.031,
      "step": 766
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.2571,
      "step": 767
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.5423,
      "step": 768
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.3712,
      "step": 769
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.9461,
      "step": 770
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.7638,
      "step": 771
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.184,
      "step": 772
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 1.0593,
      "step": 773
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0002,
      "loss": 1.1542,
      "step": 774
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.169,
      "step": 775
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.2097,
      "step": 776
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 1.289,
      "step": 777
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 1.2909,
      "step": 778
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.963,
      "step": 779
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 0.4327,
      "step": 780
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.2966,
      "step": 781
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.1678,
      "step": 782
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.3882,
      "step": 783
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.2642,
      "step": 784
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2435,
      "step": 785
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.4885,
      "step": 786
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0002,
      "loss": 1.2197,
      "step": 787
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.0192,
      "step": 788
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.1973,
      "step": 789
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.956,
      "step": 790
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.3179,
      "step": 791
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 0.8112,
      "step": 792
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.6738,
      "step": 793
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.3215,
      "step": 794
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.9692,
      "step": 795
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.0649,
      "step": 796
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.9791,
      "step": 797
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.3862,
      "step": 798
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 0.8109,
      "step": 799
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.1606,
      "step": 800
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.411,
      "step": 801
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 0.4556,
      "step": 802
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 0.7827,
      "step": 803
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.2944,
      "step": 804
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.8947,
      "step": 805
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.2127,
      "step": 806
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.0113,
      "step": 807
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 0.859,
      "step": 808
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.4038,
      "step": 809
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0002,
      "loss": 0.7468,
      "step": 810
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.2652,
      "step": 811
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.3932,
      "step": 812
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.0554,
      "step": 813
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 1.0628,
      "step": 814
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1322,
      "step": 815
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.1309,
      "step": 816
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.3735,
      "step": 817
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 0.8485,
      "step": 818
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.0253,
      "step": 819
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.3239,
      "step": 820
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.5294,
      "step": 821
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.304,
      "step": 822
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.3952,
      "step": 823
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.2594,
      "step": 824
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.0376,
      "step": 825
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.1188,
      "step": 826
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.2655,
      "step": 827
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.7659,
      "step": 828
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.2561,
      "step": 829
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3231,
      "step": 830
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0002,
      "loss": 0.9735,
      "step": 831
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.1705,
      "step": 832
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.2054,
      "step": 833
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.1353,
      "step": 834
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.0134,
      "step": 835
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.0437,
      "step": 836
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.1485,
      "step": 837
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1867,
      "step": 838
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 1.0093,
      "step": 839
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 0.7263,
      "step": 840
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.0004,
      "step": 841
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.0202,
      "step": 842
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0854,
      "step": 843
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.3441,
      "step": 844
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2956,
      "step": 845
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 1.0229,
      "step": 846
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 1.108,
      "step": 847
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.9101,
      "step": 848
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 0.7816,
      "step": 849
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 0.8355,
      "step": 850
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.2637,
      "step": 851
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.1806,
      "step": 852
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.3785,
      "step": 853
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.3641,
      "step": 854
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.134,
      "step": 855
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.2511,
      "step": 856
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.2473,
      "step": 857
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0002,
      "loss": 1.0579,
      "step": 858
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.0283,
      "step": 859
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.0823,
      "step": 860
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.3792,
      "step": 861
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.1463,
      "step": 862
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 0.8207,
      "step": 863
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0002,
      "loss": 0.8407,
      "step": 864
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 0.997,
      "step": 865
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.2884,
      "step": 866
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.268,
      "step": 867
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.004,
      "step": 868
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.1793,
      "step": 869
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.0087,
      "step": 870
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.2295,
      "step": 871
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.1312,
      "step": 872
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.4446,
      "step": 873
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.1248,
      "step": 874
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.3515,
      "step": 875
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.2335,
      "step": 876
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.4844,
      "step": 877
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 0.9647,
      "step": 878
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.2085,
      "step": 879
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.0271,
      "step": 880
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.2109,
      "step": 881
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2139,
      "step": 882
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.1757,
      "step": 883
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.9626,
      "step": 884
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 1.0561,
      "step": 885
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.0718,
      "step": 886
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.0024,
      "step": 887
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 0.8309,
      "step": 888
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.072,
      "step": 889
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.2128,
      "step": 890
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.4951,
      "step": 891
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.8999,
      "step": 892
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 1.309,
      "step": 893
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.0659,
      "step": 894
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.7715,
      "step": 895
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 0.997,
      "step": 896
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 1.2807,
      "step": 897
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.2596,
      "step": 898
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.0154,
      "step": 899
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.2994,
      "step": 900
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.1724,
      "step": 901
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.059814453125,
      "learning_rate": 0.0002,
      "loss": 0.546,
      "step": 902
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.3952,
      "step": 903
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0002,
      "loss": 0.7911,
      "step": 904
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0002,
      "loss": 1.2575,
      "step": 905
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.0513,
      "step": 906
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 1.1888,
      "step": 907
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.0626,
      "step": 908
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 0.7788,
      "step": 909
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0002,
      "loss": 0.8657,
      "step": 910
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 1.1614,
      "step": 911
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0002,
      "loss": 1.0921,
      "step": 912
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.2117,
      "step": 913
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.4573,
      "step": 914
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.2471,
      "step": 915
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0178,
      "step": 916
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.2785,
      "step": 917
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.1759,
      "step": 918
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.5018,
      "step": 919
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 1.2128,
      "step": 920
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 0.9776,
      "step": 921
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.1665,
      "step": 922
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.7345,
      "step": 923
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.0715,
      "step": 924
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.0826,
      "step": 925
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.71875,
      "learning_rate": 0.0002,
      "loss": 1.6547,
      "step": 926
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.8527,
      "step": 927
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.3448,
      "step": 928
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.4396,
      "step": 929
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.1593,
      "step": 930
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.3491,
      "step": 931
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.1137,
      "step": 932
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.1548,
      "step": 933
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.0642,
      "step": 934
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.4708,
      "step": 935
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2116,
      "step": 936
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.0253,
      "step": 937
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.9823,
      "step": 938
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 0.7361,
      "step": 939
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.9492,
      "step": 940
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.0841,
      "step": 941
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.1289,
      "step": 942
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.4103,
      "step": 943
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.372,
      "step": 944
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0002,
      "loss": 0.8263,
      "step": 945
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0002,
      "loss": 1.541,
      "step": 946
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.2887,
      "step": 947
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.3515,
      "step": 948
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.0866,
      "step": 949
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 1.4229,
      "step": 950
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.2345,
      "step": 951
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.3922,
      "step": 952
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.3256,
      "step": 953
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.0828,
      "step": 954
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.188,
      "step": 955
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.9938,
      "step": 956
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2858,
      "step": 957
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 1.0628,
      "step": 958
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 0.8584,
      "step": 959
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.0627,
      "step": 960
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1486,
      "step": 961
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.7363,
      "step": 962
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.0196,
      "step": 963
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.362,
      "step": 964
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.2739,
      "step": 965
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.2585,
      "step": 966
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.1658,
      "step": 967
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.7626,
      "step": 968
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0771,
      "step": 969
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.2686,
      "step": 970
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.1938,
      "step": 971
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0002,
      "loss": 0.6803,
      "step": 972
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 0.8585,
      "step": 973
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.9559,
      "step": 974
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.4466,
      "step": 975
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.1558,
      "step": 976
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 0.99,
      "step": 977
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.1851,
      "step": 978
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.1278,
      "step": 979
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.2371,
      "step": 980
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2097,
      "step": 981
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.0208,
      "step": 982
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.1149,
      "step": 983
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 0.8109,
      "step": 984
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.2963,
      "step": 985
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 0.7968,
      "step": 986
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0002,
      "loss": 1.0684,
      "step": 987
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 0.9528,
      "step": 988
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.2903,
      "step": 989
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.3793,
      "step": 990
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.9861,
      "step": 991
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.1672,
      "step": 992
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.4268,
      "step": 993
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.5739,
      "step": 994
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0002,
      "loss": 0.9054,
      "step": 995
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1098,
      "step": 996
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 1.0959,
      "step": 997
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.072265625,
      "learning_rate": 0.0002,
      "loss": 0.5114,
      "step": 998
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.5086,
      "step": 999
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.2697,
      "step": 1000
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 0.9952,
      "step": 1001
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.8092,
      "step": 1002
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 1.033,
      "step": 1003
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 0.8184,
      "step": 1004
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.2774,
      "step": 1005
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 0.7665,
      "step": 1006
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.13,
      "step": 1007
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0002,
      "loss": 0.7389,
      "step": 1008
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.3348,
      "step": 1009
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.1257,
      "step": 1010
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.1226,
      "step": 1011
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.1329,
      "step": 1012
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 1.2394,
      "step": 1013
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.3612,
      "step": 1014
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.0114,
      "step": 1015
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.4029,
      "step": 1016
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.3795,
      "step": 1017
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.9664,
      "step": 1018
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.9079,
      "step": 1019
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.6494,
      "step": 1020
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.2784,
      "step": 1021
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.0038,
      "step": 1022
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.3129,
      "step": 1023
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.2789,
      "step": 1024
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.3669,
      "step": 1025
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.0354,
      "step": 1026
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.3622,
      "step": 1027
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.9071,
      "step": 1028
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.0115,
      "step": 1029
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.4787,
      "step": 1030
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.1625,
      "step": 1031
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 1.1425,
      "step": 1032
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.2003,
      "step": 1033
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.0465,
      "step": 1034
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.4793,
      "step": 1035
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1978,
      "step": 1036
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.5347,
      "step": 1037
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 1.0631,
      "step": 1038
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.9241,
      "step": 1039
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.1041,
      "step": 1040
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.0786,
      "step": 1041
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0002,
      "loss": 0.9614,
      "step": 1042
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.2478,
      "step": 1043
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.9375,
      "step": 1044
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 0.9722,
      "step": 1045
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 0.6146,
      "step": 1046
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.7556,
      "step": 1047
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 0.9992,
      "step": 1048
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 0.8592,
      "step": 1049
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.0334,
      "step": 1050
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.0233,
      "step": 1051
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.4049,
      "step": 1052
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.1988,
      "step": 1053
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.7725,
      "step": 1054
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0002,
      "loss": 1.2065,
      "step": 1055
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.4533,
      "step": 1056
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2756,
      "step": 1057
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.0054,
      "step": 1058
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.7157,
      "step": 1059
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 1.1692,
      "step": 1060
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.1324,
      "step": 1061
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.1972,
      "step": 1062
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.5428,
      "step": 1063
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.2159,
      "step": 1064
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.9314,
      "step": 1065
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.2579,
      "step": 1066
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.1103,
      "step": 1067
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.0554,
      "step": 1068
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 1.1991,
      "step": 1069
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.9415,
      "step": 1070
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.311,
      "step": 1071
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 0.9924,
      "step": 1072
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.0061,
      "step": 1073
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 0.8137,
      "step": 1074
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0228,
      "step": 1075
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.4037,
      "step": 1076
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.1407,
      "step": 1077
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 0.6942,
      "step": 1078
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.4618,
      "step": 1079
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.3025,
      "step": 1080
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.214,
      "step": 1081
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.5483,
      "step": 1082
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 0.4809,
      "step": 1083
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.0334,
      "step": 1084
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.3541,
      "step": 1085
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 0.9011,
      "step": 1086
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0002,
      "loss": 0.8841,
      "step": 1087
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.1744,
      "step": 1088
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0002,
      "loss": 0.8115,
      "step": 1089
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.0467,
      "step": 1090
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.1719,
      "step": 1091
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.2096,
      "step": 1092
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.3748,
      "step": 1093
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.099,
      "step": 1094
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.6086,
      "step": 1095
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0002,
      "loss": 1.381,
      "step": 1096
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.3321,
      "step": 1097
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.1062,
      "step": 1098
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.3739,
      "step": 1099
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 0.9878,
      "step": 1100
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.3442,
      "step": 1101
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.2821,
      "step": 1102
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.0409,
      "step": 1103
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0224,
      "step": 1104
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.213,
      "step": 1105
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 0.8655,
      "step": 1106
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 0.9087,
      "step": 1107
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 0.7064,
      "step": 1108
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.254,
      "step": 1109
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.0624,
      "step": 1110
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.8707,
      "step": 1111
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.113,
      "step": 1112
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.3672,
      "step": 1113
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.1728,
      "step": 1114
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.0153,
      "step": 1115
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 0.9375,
      "step": 1116
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.7338,
      "step": 1117
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.0765,
      "step": 1118
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.1908,
      "step": 1119
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.2139,
      "step": 1120
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.2155,
      "step": 1121
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.4918,
      "step": 1122
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.149,
      "step": 1123
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.7929,
      "step": 1124
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.082,
      "step": 1125
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 1.4693,
      "step": 1126
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.5901,
      "step": 1127
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.6763,
      "step": 1128
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.1515,
      "step": 1129
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 0.9948,
      "step": 1130
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.5467,
      "step": 1131
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0002,
      "loss": 1.3228,
      "step": 1132
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.3067,
      "step": 1133
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 0.9549,
      "step": 1134
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0002,
      "loss": 1.0922,
      "step": 1135
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.3987,
      "step": 1136
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.2654,
      "step": 1137
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 1.1451,
      "step": 1138
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0002,
      "loss": 1.1085,
      "step": 1139
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.3173,
      "step": 1140
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 1.0928,
      "step": 1141
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 0.7343,
      "step": 1142
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 0.9903,
      "step": 1143
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0002,
      "loss": 1.0577,
      "step": 1144
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.1098,
      "step": 1145
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 0.915,
      "step": 1146
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.3625,
      "step": 1147
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 0.9699,
      "step": 1148
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.5113,
      "step": 1149
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.2319,
      "step": 1150
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.8574,
      "step": 1151
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.6273,
      "step": 1152
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.1519,
      "step": 1153
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.1047,
      "step": 1154
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.0007,
      "step": 1155
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.4835,
      "step": 1156
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 0.9935,
      "step": 1157
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.2896,
      "step": 1158
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.3761,
      "step": 1159
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.0727,
      "step": 1160
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.1636,
      "step": 1161
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.6392,
      "step": 1162
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.4179,
      "step": 1163
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.2881,
      "step": 1164
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.1229,
      "step": 1165
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.9964,
      "step": 1166
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 1.3119,
      "step": 1167
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.3433,
      "step": 1168
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.8222,
      "step": 1169
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.1417,
      "step": 1170
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.8481,
      "step": 1171
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.2496,
      "step": 1172
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.4369,
      "step": 1173
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.9684,
      "step": 1174
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.4815,
      "step": 1175
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0002,
      "loss": 1.3405,
      "step": 1176
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.9888,
      "step": 1177
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3094,
      "step": 1178
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0002,
      "loss": 0.9803,
      "step": 1179
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.0683,
      "step": 1180
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1889,
      "step": 1181
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.3423,
      "step": 1182
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.4883,
      "step": 1183
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.0919,
      "step": 1184
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 1.1173,
      "step": 1185
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.3227,
      "step": 1186
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 1.2167,
      "step": 1187
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0002,
      "loss": 1.1635,
      "step": 1188
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.3145,
      "step": 1189
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.9408,
      "step": 1190
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.2477,
      "step": 1191
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.0281,
      "step": 1192
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.1536,
      "step": 1193
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.0565,
      "step": 1194
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 1.2036,
      "step": 1195
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 0.9779,
      "step": 1196
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.3977,
      "step": 1197
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 1.0063,
      "step": 1198
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.0221,
      "step": 1199
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.2519,
      "step": 1200
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.2882,
      "step": 1201
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.909,
      "step": 1202
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.0992,
      "step": 1203
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1833,
      "step": 1204
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.1,
      "step": 1205
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 0.8045,
      "step": 1206
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.0861,
      "step": 1207
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.413,
      "step": 1208
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.2464,
      "step": 1209
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 0.8169,
      "step": 1210
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.0237,
      "step": 1211
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.3505,
      "step": 1212
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 0.8023,
      "step": 1213
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.1201,
      "step": 1214
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.2701,
      "step": 1215
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.3293,
      "step": 1216
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.9447,
      "step": 1217
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.1986,
      "step": 1218
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.2968,
      "step": 1219
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0858,
      "step": 1220
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.3434,
      "step": 1221
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.2316,
      "step": 1222
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 1.2379,
      "step": 1223
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.6175,
      "step": 1224
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 0.907,
      "step": 1225
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.1844,
      "step": 1226
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.3519,
      "step": 1227
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 0.8523,
      "step": 1228
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.202,
      "step": 1229
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.3678,
      "step": 1230
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 1.1214,
      "step": 1231
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.1557,
      "step": 1232
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.2614,
      "step": 1233
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.0853,
      "step": 1234
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 0.8999,
      "step": 1235
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 0.6993,
      "step": 1236
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.204,
      "step": 1237
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.0611,
      "step": 1238
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.0411,
      "step": 1239
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.1006,
      "step": 1240
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.1758,
      "step": 1241
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.1684,
      "step": 1242
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.7137,
      "step": 1243
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0933,
      "step": 1244
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.4282,
      "step": 1245
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.3429,
      "step": 1246
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.0424,
      "step": 1247
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.017,
      "step": 1248
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.0223,
      "step": 1249
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.1696,
      "step": 1250
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.2318,
      "step": 1251
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.0361,
      "step": 1252
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.3758,
      "step": 1253
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 0.9246,
      "step": 1254
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0002,
      "loss": 0.9803,
      "step": 1255
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.1382,
      "step": 1256
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.0347,
      "step": 1257
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 0.947,
      "step": 1258
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 0.4067,
      "step": 1259
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.0806,
      "step": 1260
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.9618,
      "step": 1261
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 0.7979,
      "step": 1262
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.1765,
      "step": 1263
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0002,
      "loss": 1.1962,
      "step": 1264
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 0.8787,
      "step": 1265
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 0.9255,
      "step": 1266
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 0.8099,
      "step": 1267
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.4374,
      "step": 1268
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.0937,
      "step": 1269
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.2819,
      "step": 1270
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0002,
      "loss": 1.3307,
      "step": 1271
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.7177,
      "step": 1272
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.9754,
      "step": 1273
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.0676,
      "step": 1274
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 0.6678,
      "step": 1275
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 0.9335,
      "step": 1276
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.3954,
      "step": 1277
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.0913,
      "step": 1278
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.9247,
      "step": 1279
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 1.1561,
      "step": 1280
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.4627,
      "step": 1281
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.1369,
      "step": 1282
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.2387,
      "step": 1283
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0002,
      "loss": 1.1376,
      "step": 1284
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0002,
      "loss": 1.3114,
      "step": 1285
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0002,
      "loss": 1.0921,
      "step": 1286
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.0598,
      "step": 1287
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.4461,
      "step": 1288
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.0309,
      "step": 1289
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.0919,
      "step": 1290
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 1.2482,
      "step": 1291
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.0955,
      "step": 1292
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 0.9735,
      "step": 1293
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0002,
      "loss": 1.0384,
      "step": 1294
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.1166,
      "step": 1295
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.2031,
      "step": 1296
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.1754,
      "step": 1297
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.0511,
      "step": 1298
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 0.6403,
      "step": 1299
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 0.9883,
      "step": 1300
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.9851,
      "step": 1301
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.1318,
      "step": 1302
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.9796,
      "step": 1303
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.8123,
      "step": 1304
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 0.988,
      "step": 1305
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0002,
      "loss": 1.0201,
      "step": 1306
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0002,
      "loss": 1.1174,
      "step": 1307
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.2495,
      "step": 1308
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 0.9503,
      "step": 1309
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0002,
      "loss": 0.7224,
      "step": 1310
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.181,
      "step": 1311
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 0.8448,
      "step": 1312
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 1.0908,
      "step": 1313
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 0.846,
      "step": 1314
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.0275,
      "step": 1315
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.3504,
      "step": 1316
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.0205,
      "step": 1317
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 0.7949,
      "step": 1318
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.6269,
      "step": 1319
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.0083,
      "step": 1320
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.1008,
      "step": 1321
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.7007,
      "step": 1322
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.0972,
      "step": 1323
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.3115,
      "step": 1324
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 0.9147,
      "step": 1325
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0002,
      "loss": 0.8808,
      "step": 1326
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 0.7656,
      "step": 1327
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.3929,
      "step": 1328
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.4706,
      "step": 1329
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.2282,
      "step": 1330
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.095,
      "step": 1331
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 0.9801,
      "step": 1332
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.9854,
      "step": 1333
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.2175,
      "step": 1334
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.1,
      "step": 1335
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.7447,
      "step": 1336
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.362,
      "step": 1337
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.8338,
      "step": 1338
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.0324,
      "step": 1339
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.0024,
      "step": 1340
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.2062,
      "step": 1341
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 1.1549,
      "step": 1342
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.4042,
      "step": 1343
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.2255,
      "step": 1344
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.3979,
      "step": 1345
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.1037,
      "step": 1346
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0002,
      "loss": 1.3351,
      "step": 1347
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.4012,
      "step": 1348
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.0717,
      "step": 1349
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0872,
      "step": 1350
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.8791,
      "step": 1351
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.1407,
      "step": 1352
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.4987,
      "step": 1353
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 0.8318,
      "step": 1354
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.384,
      "step": 1355
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0002,
      "loss": 1.102,
      "step": 1356
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 0.9981,
      "step": 1357
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.8926,
      "step": 1358
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0002,
      "loss": 0.8748,
      "step": 1359
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.2755,
      "step": 1360
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 1.1515,
      "step": 1361
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.3589,
      "step": 1362
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.249,
      "step": 1363
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 0.7953,
      "step": 1364
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.0454,
      "step": 1365
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.1191,
      "step": 1366
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 0.968,
      "step": 1367
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.0048,
      "step": 1368
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0002,
      "loss": 1.1549,
      "step": 1369
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.4864,
      "step": 1370
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.5424,
      "step": 1371
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.4756,
      "step": 1372
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.9445,
      "step": 1373
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.078,
      "step": 1374
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.4644,
      "step": 1375
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.5975,
      "step": 1376
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.2136,
      "step": 1377
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2584,
      "step": 1378
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 0.8689,
      "step": 1379
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.1057,
      "step": 1380
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.0443,
      "step": 1381
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0002,
      "loss": 1.0637,
      "step": 1382
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.3618,
      "step": 1383
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.3944,
      "step": 1384
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 1.0997,
      "step": 1385
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.8815,
      "step": 1386
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0002,
      "loss": 0.9601,
      "step": 1387
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.0734,
      "step": 1388
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.5368,
      "step": 1389
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 0.9998,
      "step": 1390
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.9045,
      "step": 1391
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 0.7854,
      "step": 1392
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 0.9518,
      "step": 1393
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.0406,
      "step": 1394
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.2177,
      "step": 1395
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0555,
      "step": 1396
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.9517,
      "step": 1397
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.0315,
      "step": 1398
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.8835,
      "step": 1399
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.0761,
      "step": 1400
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 0.9252,
      "step": 1401
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.2525,
      "step": 1402
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.976,
      "step": 1403
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.6801,
      "step": 1404
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.1881,
      "step": 1405
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.0155,
      "step": 1406
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.3532,
      "step": 1407
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.219,
      "step": 1408
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.3726,
      "step": 1409
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.9789,
      "step": 1410
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 0.897,
      "step": 1411
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3902,
      "step": 1412
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.0912,
      "step": 1413
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 0.6696,
      "step": 1414
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.0134,
      "step": 1415
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.9131,
      "step": 1416
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.1802,
      "step": 1417
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0002,
      "loss": 0.6913,
      "step": 1418
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.6722,
      "step": 1419
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.3054,
      "step": 1420
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.4818,
      "step": 1421
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.3833,
      "step": 1422
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.0582,
      "step": 1423
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.3072,
      "step": 1424
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.1437,
      "step": 1425
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.0741,
      "step": 1426
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.8846,
      "step": 1427
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.3554,
      "step": 1428
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1733,
      "step": 1429
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.2234,
      "step": 1430
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.0591,
      "step": 1431
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.9115,
      "step": 1432
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.6669,
      "step": 1433
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.248,
      "step": 1434
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.4346,
      "step": 1435
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.6415,
      "step": 1436
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.0427,
      "step": 1437
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.1837,
      "step": 1438
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 2.1509,
      "step": 1439
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.2885,
      "step": 1440
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.1479,
      "step": 1441
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.2692,
      "step": 1442
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.339,
      "step": 1443
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.3213,
      "step": 1444
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.3065,
      "step": 1445
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.8563,
      "step": 1446
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.4148,
      "step": 1447
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.2823,
      "step": 1448
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0002,
      "loss": 1.1107,
      "step": 1449
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.2343,
      "step": 1450
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.4031,
      "step": 1451
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 0.8841,
      "step": 1452
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.3564,
      "step": 1453
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.1623,
      "step": 1454
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.2627,
      "step": 1455
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 1.1454,
      "step": 1456
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.1488,
      "step": 1457
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2344,
      "step": 1458
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 1.0374,
      "step": 1459
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.0856,
      "step": 1460
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.2479,
      "step": 1461
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 0.9165,
      "step": 1462
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 0.9895,
      "step": 1463
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 1.2038,
      "step": 1464
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 0.9444,
      "step": 1465
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.5476,
      "step": 1466
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.3025,
      "step": 1467
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.0314,
      "step": 1468
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.4499,
      "step": 1469
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.9352,
      "step": 1470
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.8292,
      "step": 1471
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0784,
      "step": 1472
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.9,
      "step": 1473
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.0464,
      "step": 1474
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.0789,
      "step": 1475
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.2908,
      "step": 1476
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1335,
      "step": 1477
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.9772,
      "step": 1478
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.2105,
      "step": 1479
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.1932,
      "step": 1480
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.0474,
      "step": 1481
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.1458,
      "step": 1482
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0002,
      "loss": 1.0252,
      "step": 1483
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.4116,
      "step": 1484
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.1256,
      "step": 1485
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.2858,
      "step": 1486
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.0206,
      "step": 1487
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.8919,
      "step": 1488
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 0.8734,
      "step": 1489
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.3375,
      "step": 1490
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.2563,
      "step": 1491
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.7159,
      "step": 1492
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 0.8056,
      "step": 1493
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.9092,
      "step": 1494
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 1.4426,
      "step": 1495
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 1.1885,
      "step": 1496
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.8223,
      "step": 1497
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.316,
      "step": 1498
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.1138,
      "step": 1499
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.0886,
      "step": 1500
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0002,
      "loss": 0.595,
      "step": 1501
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.7876,
      "step": 1502
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.2187,
      "step": 1503
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.08,
      "step": 1504
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.2859,
      "step": 1505
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.0019,
      "step": 1506
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0002,
      "loss": 1.0439,
      "step": 1507
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.9991,
      "step": 1508
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.8998,
      "step": 1509
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.5935,
      "step": 1510
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.9289,
      "step": 1511
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 0.8891,
      "step": 1512
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.9671,
      "step": 1513
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2408,
      "step": 1514
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.8699,
      "step": 1515
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.241,
      "step": 1516
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3649,
      "step": 1517
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.3664,
      "step": 1518
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 0.5985,
      "step": 1519
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.9861,
      "step": 1520
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.1461,
      "step": 1521
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 0.9997,
      "step": 1522
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.319,
      "step": 1523
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.011,
      "step": 1524
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.416,
      "step": 1525
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.1253,
      "step": 1526
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.0799,
      "step": 1527
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.1636,
      "step": 1528
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 0.8677,
      "step": 1529
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0002,
      "loss": 1.1499,
      "step": 1530
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.2899,
      "step": 1531
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.9754,
      "step": 1532
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.0264,
      "step": 1533
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.8915,
      "step": 1534
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.0652,
      "step": 1535
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0002,
      "loss": 1.1082,
      "step": 1536
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 1.084,
      "step": 1537
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.4419,
      "step": 1538
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0002,
      "loss": 0.9287,
      "step": 1539
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 0.9028,
      "step": 1540
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.3604,
      "step": 1541
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.1209,
      "step": 1542
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.9448,
      "step": 1543
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.8781,
      "step": 1544
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 0.9276,
      "step": 1545
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0002,
      "loss": 0.9274,
      "step": 1546
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0002,
      "loss": 1.2839,
      "step": 1547
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.3672,
      "step": 1548
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.1058,
      "step": 1549
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 1.2261,
      "step": 1550
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.9064,
      "step": 1551
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.4516,
      "step": 1552
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.627,
      "step": 1553
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.1177,
      "step": 1554
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0002,
      "loss": 1.0632,
      "step": 1555
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.1505,
      "step": 1556
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.2835,
      "step": 1557
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.0735,
      "step": 1558
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.9865,
      "step": 1559
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0002,
      "loss": 0.9211,
      "step": 1560
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.3859,
      "step": 1561
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 0.8464,
      "step": 1562
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 1.1953,
      "step": 1563
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 0.8259,
      "step": 1564
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 1.0463,
      "step": 1565
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.1747,
      "step": 1566
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2034,
      "step": 1567
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.0537,
      "step": 1568
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.0479,
      "step": 1569
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0002,
      "loss": 1.0208,
      "step": 1570
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 1.0646,
      "step": 1571
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.4312,
      "step": 1572
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 1.1898,
      "step": 1573
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 0.989,
      "step": 1574
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.1244,
      "step": 1575
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.0446,
      "step": 1576
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.3168,
      "step": 1577
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 1.0485,
      "step": 1578
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 0.9811,
      "step": 1579
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.1911,
      "step": 1580
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.9585,
      "step": 1581
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0002,
      "loss": 1.0037,
      "step": 1582
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1609,
      "step": 1583
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0103,
      "step": 1584
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.5713,
      "step": 1585
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.3266,
      "step": 1586
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.0746,
      "step": 1587
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 0.8853,
      "step": 1588
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.0539,
      "step": 1589
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.2494,
      "step": 1590
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.8899,
      "step": 1591
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.2113,
      "step": 1592
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.8741,
      "step": 1593
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 1.2849,
      "step": 1594
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.1358,
      "step": 1595
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.078125,
      "learning_rate": 0.0002,
      "loss": 1.0904,
      "step": 1596
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.3183,
      "step": 1597
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.0627,
      "step": 1598
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.3036,
      "step": 1599
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.9895,
      "step": 1600
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.7529,
      "step": 1601
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.103,
      "step": 1602
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.0476,
      "step": 1603
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.0571,
      "step": 1604
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0002,
      "loss": 0.4966,
      "step": 1605
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0002,
      "loss": 1.1147,
      "step": 1606
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 0.753,
      "step": 1607
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.1006,
      "step": 1608
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.0185,
      "step": 1609
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 0.7396,
      "step": 1610
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.4387,
      "step": 1611
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 0.761,
      "step": 1612
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.7068,
      "step": 1613
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0385,
      "step": 1614
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 0.474,
      "step": 1615
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.0684,
      "step": 1616
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 0.8345,
      "step": 1617
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.6335,
      "step": 1618
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.19,
      "step": 1619
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.336,
      "step": 1620
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.3921,
      "step": 1621
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.2693,
      "step": 1622
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0002,
      "loss": 0.9578,
      "step": 1623
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 1.0992,
      "step": 1624
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.0734,
      "step": 1625
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.0451,
      "step": 1626
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.3788,
      "step": 1627
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 1.0532,
      "step": 1628
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.2564,
      "step": 1629
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.1286,
      "step": 1630
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0002,
      "loss": 0.8477,
      "step": 1631
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.6682,
      "step": 1632
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.0197,
      "step": 1633
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.403,
      "step": 1634
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.5666,
      "step": 1635
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.2723,
      "step": 1636
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 0.4126,
      "step": 1637
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.1727,
      "step": 1638
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.3479,
      "step": 1639
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.1233,
      "step": 1640
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.2041,
      "step": 1641
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.4119,
      "step": 1642
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 0.9868,
      "step": 1643
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.056,
      "step": 1644
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.2626,
      "step": 1645
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.2672,
      "step": 1646
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.080078125,
      "learning_rate": 0.0002,
      "loss": 0.4905,
      "step": 1647
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0941,
      "step": 1648
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.1919,
      "step": 1649
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 0.8929,
      "step": 1650
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1527,
      "step": 1651
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0002,
      "loss": 0.6288,
      "step": 1652
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.2144,
      "step": 1653
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.2423,
      "step": 1654
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 1.0952,
      "step": 1655
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.3988,
      "step": 1656
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.1446,
      "step": 1657
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.2537,
      "step": 1658
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.0731,
      "step": 1659
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 1.0612,
      "step": 1660
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.0336,
      "step": 1661
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.2096,
      "step": 1662
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.0421,
      "step": 1663
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0101,
      "step": 1664
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 1.3059,
      "step": 1665
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.9007,
      "step": 1666
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 1.2102,
      "step": 1667
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.1662,
      "step": 1668
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.0314,
      "step": 1669
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.6331,
      "step": 1670
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.0408,
      "step": 1671
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.0826,
      "step": 1672
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.1371,
      "step": 1673
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.0653,
      "step": 1674
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.0264,
      "step": 1675
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.1547,
      "step": 1676
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 0.4176,
      "step": 1677
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.0703,
      "step": 1678
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.2297,
      "step": 1679
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.4202,
      "step": 1680
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2716,
      "step": 1681
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.0126,
      "step": 1682
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.1905,
      "step": 1683
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 0.9118,
      "step": 1684
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.3563,
      "step": 1685
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.1564,
      "step": 1686
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.3431,
      "step": 1687
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.0725,
      "step": 1688
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.3381,
      "step": 1689
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.1551,
      "step": 1690
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.1886,
      "step": 1691
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.3659,
      "step": 1692
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.901,
      "step": 1693
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0002,
      "loss": 1.0055,
      "step": 1694
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.9524,
      "step": 1695
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.158,
      "step": 1696
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.5352,
      "step": 1697
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.7418,
      "step": 1698
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.2509,
      "step": 1699
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.1228,
      "step": 1700
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.1393,
      "step": 1701
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.0344,
      "step": 1702
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.4517,
      "step": 1703
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.2241,
      "step": 1704
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 1.3348,
      "step": 1705
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.2891,
      "step": 1706
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.5463,
      "step": 1707
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 1.0454,
      "step": 1708
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 0.8657,
      "step": 1709
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.2814,
      "step": 1710
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.5872,
      "step": 1711
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.3115,
      "step": 1712
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 0.7695,
      "step": 1713
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2304,
      "step": 1714
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0002,
      "loss": 0.9379,
      "step": 1715
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 0.9915,
      "step": 1716
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.3484,
      "step": 1717
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 0.4492,
      "step": 1718
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.2622,
      "step": 1719
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.0954,
      "step": 1720
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.0378,
      "step": 1721
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.8796,
      "step": 1722
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 0.8912,
      "step": 1723
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.3479,
      "step": 1724
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 1.312,
      "step": 1725
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.2979,
      "step": 1726
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0002,
      "loss": 0.9941,
      "step": 1727
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3172,
      "step": 1728
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 0.5152,
      "step": 1729
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 1.169,
      "step": 1730
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.09375,
      "learning_rate": 0.0002,
      "loss": 0.8299,
      "step": 1731
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 0.3423,
      "step": 1732
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.4003,
      "step": 1733
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.2733,
      "step": 1734
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.6601,
      "step": 1735
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 0.9626,
      "step": 1736
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 0.7304,
      "step": 1737
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.2294,
      "step": 1738
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.2702,
      "step": 1739
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.0857,
      "step": 1740
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 0.8282,
      "step": 1741
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.0334,
      "step": 1742
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.2441,
      "step": 1743
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 1.0968,
      "step": 1744
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.9252,
      "step": 1745
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.0504,
      "step": 1746
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.199,
      "step": 1747
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 0.8798,
      "step": 1748
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.516,
      "step": 1749
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.2046,
      "step": 1750
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0002,
      "loss": 1.3222,
      "step": 1751
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.07,
      "step": 1752
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.3057,
      "step": 1753
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.3071,
      "step": 1754
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.3611,
      "step": 1755
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.288,
      "step": 1756
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.0573,
      "step": 1757
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.9681,
      "step": 1758
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.9904,
      "step": 1759
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.0458,
      "step": 1760
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.3919,
      "step": 1761
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.1572,
      "step": 1762
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1325,
      "step": 1763
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.2402,
      "step": 1764
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 0.4755,
      "step": 1765
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0002,
      "loss": 1.0111,
      "step": 1766
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 1.0744,
      "step": 1767
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 1.7494,
      "step": 1768
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.103,
      "step": 1769
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.4122,
      "step": 1770
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.201,
      "step": 1771
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.3496,
      "step": 1772
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.7772,
      "step": 1773
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.1722,
      "step": 1774
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.058,
      "step": 1775
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.2407,
      "step": 1776
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2779,
      "step": 1777
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.9599,
      "step": 1778
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.1842,
      "step": 1779
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.4624,
      "step": 1780
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.2417,
      "step": 1781
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.1635,
      "step": 1782
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.09033203125,
      "learning_rate": 0.0002,
      "loss": 0.5821,
      "step": 1783
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0002,
      "loss": 1.3416,
      "step": 1784
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.9979,
      "step": 1785
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0002,
      "loss": 1.1497,
      "step": 1786
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.0402,
      "step": 1787
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 1.0132,
      "step": 1788
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.6958,
      "step": 1789
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.197,
      "step": 1790
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.2606,
      "step": 1791
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.4572,
      "step": 1792
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 1.1195,
      "step": 1793
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.0736,
      "step": 1794
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.0147,
      "step": 1795
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.1162,
      "step": 1796
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.6785,
      "step": 1797
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.2507,
      "step": 1798
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 0.8843,
      "step": 1799
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.0301,
      "step": 1800
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0002,
      "loss": 0.8679,
      "step": 1801
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.0939,
      "step": 1802
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.2816,
      "step": 1803
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.9675,
      "step": 1804
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.9514,
      "step": 1805
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.9354,
      "step": 1806
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.301,
      "step": 1807
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0002,
      "loss": 0.9781,
      "step": 1808
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.2797,
      "step": 1809
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 1.0722,
      "step": 1810
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 0.9621,
      "step": 1811
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.1868,
      "step": 1812
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.3577,
      "step": 1813
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.1715,
      "step": 1814
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0002,
      "loss": 0.7861,
      "step": 1815
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.5165,
      "step": 1816
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0002,
      "loss": 0.6334,
      "step": 1817
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.2814,
      "step": 1818
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 0.7443,
      "step": 1819
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.1638,
      "step": 1820
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.2638,
      "step": 1821
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.9023,
      "step": 1822
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.0019,
      "step": 1823
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0607,
      "step": 1824
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1966,
      "step": 1825
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 0.7891,
      "step": 1826
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.168,
      "step": 1827
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.1994,
      "step": 1828
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.0594,
      "step": 1829
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.2487,
      "step": 1830
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.0309,
      "step": 1831
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0002,
      "loss": 1.2767,
      "step": 1832
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.9348,
      "step": 1833
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.1173,
      "step": 1834
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 1.2275,
      "step": 1835
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.7517,
      "step": 1836
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.2981,
      "step": 1837
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.3845,
      "step": 1838
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.8863,
      "step": 1839
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0002,
      "loss": 0.7421,
      "step": 1840
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 1.3228,
      "step": 1841
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.0259,
      "step": 1842
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.2301,
      "step": 1843
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 1.2726,
      "step": 1844
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0002,
      "loss": 0.3496,
      "step": 1845
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.0548,
      "step": 1846
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.3393,
      "step": 1847
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.8617,
      "step": 1848
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.731,
      "step": 1849
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 1.0,
      "step": 1850
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.9543,
      "step": 1851
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0025,
      "step": 1852
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 0.6774,
      "step": 1853
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 0.8939,
      "step": 1854
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.478,
      "step": 1855
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.2617,
      "step": 1856
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.1068,
      "step": 1857
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 1.0932,
      "step": 1858
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0002,
      "loss": 1.3498,
      "step": 1859
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.2122,
      "step": 1860
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.095,
      "step": 1861
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 0.8963,
      "step": 1862
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.0174,
      "step": 1863
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.3041,
      "step": 1864
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 0.9742,
      "step": 1865
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.2263,
      "step": 1866
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 0.9299,
      "step": 1867
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.4094,
      "step": 1868
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.1187,
      "step": 1869
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.373,
      "step": 1870
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.948,
      "step": 1871
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 0.9325,
      "step": 1872
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.08349609375,
      "learning_rate": 0.0002,
      "loss": 0.6294,
      "step": 1873
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.2844,
      "step": 1874
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0002,
      "loss": 0.9072,
      "step": 1875
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.16,
      "step": 1876
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 0.8921,
      "step": 1877
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.9361,
      "step": 1878
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.5475,
      "step": 1879
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0002,
      "loss": 1.2399,
      "step": 1880
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 0.8917,
      "step": 1881
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.1626,
      "step": 1882
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.3287,
      "step": 1883
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 0.9757,
      "step": 1884
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.2283,
      "step": 1885
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2796,
      "step": 1886
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2103,
      "step": 1887
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.338,
      "step": 1888
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 1.1728,
      "step": 1889
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.0698,
      "step": 1890
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.9234,
      "step": 1891
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.2967,
      "step": 1892
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.1038,
      "step": 1893
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.3665,
      "step": 1894
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 0.9921,
      "step": 1895
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.3495,
      "step": 1896
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.0927,
      "step": 1897
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.2106,
      "step": 1898
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.119,
      "step": 1899
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.9366,
      "step": 1900
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.0762,
      "step": 1901
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.6584,
      "step": 1902
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.3579,
      "step": 1903
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.2512,
      "step": 1904
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.145,
      "step": 1905
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.129,
      "step": 1906
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.081,
      "step": 1907
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.1393,
      "step": 1908
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.2227,
      "step": 1909
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.1199,
      "step": 1910
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.0514,
      "step": 1911
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.4191,
      "step": 1912
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.2044,
      "step": 1913
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.1482,
      "step": 1914
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0002,
      "loss": 0.9571,
      "step": 1915
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 0.8059,
      "step": 1916
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.2762,
      "step": 1917
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.4892,
      "step": 1918
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.0372,
      "step": 1919
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.4236,
      "step": 1920
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.3155,
      "step": 1921
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.4745,
      "step": 1922
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.1757,
      "step": 1923
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.07470703125,
      "learning_rate": 0.0002,
      "loss": 0.5785,
      "step": 1924
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.127,
      "step": 1925
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.1232,
      "step": 1926
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.3376,
      "step": 1927
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.0919,
      "step": 1928
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 0.9706,
      "step": 1929
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.1103,
      "step": 1930
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.0941,
      "step": 1931
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0002,
      "loss": 1.3157,
      "step": 1932
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.0312,
      "step": 1933
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.7984,
      "step": 1934
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 0.9067,
      "step": 1935
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.891,
      "step": 1936
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.0097,
      "step": 1937
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.1242,
      "step": 1938
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.4655,
      "step": 1939
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.1605,
      "step": 1940
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.1279,
      "step": 1941
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.9309,
      "step": 1942
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.9488,
      "step": 1943
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.0272,
      "step": 1944
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.5163,
      "step": 1945
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.0246,
      "step": 1946
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 1.0298,
      "step": 1947
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0002,
      "loss": 1.3364,
      "step": 1948
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 0.6851,
      "step": 1949
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.1746,
      "step": 1950
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.9845,
      "step": 1951
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.2038,
      "step": 1952
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.3589,
      "step": 1953
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.3746,
      "step": 1954
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 1.3134,
      "step": 1955
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0125,
      "step": 1956
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.7946,
      "step": 1957
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.7909,
      "step": 1958
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2471,
      "step": 1959
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.279,
      "step": 1960
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.1583,
      "step": 1961
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.3243,
      "step": 1962
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.406,
      "step": 1963
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.3084,
      "step": 1964
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 1.0102,
      "step": 1965
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.0725,
      "step": 1966
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.0971,
      "step": 1967
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 1.2232,
      "step": 1968
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 0.993,
      "step": 1969
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1845,
      "step": 1970
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.2296,
      "step": 1971
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0523,
      "step": 1972
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.35,
      "step": 1973
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.215,
      "step": 1974
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0002,
      "loss": 0.9473,
      "step": 1975
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.3339,
      "step": 1976
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0204,
      "step": 1977
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3088,
      "step": 1978
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.0537,
      "step": 1979
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0424,
      "step": 1980
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.1618,
      "step": 1981
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 1.0366,
      "step": 1982
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.1349,
      "step": 1983
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.2601,
      "step": 1984
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.3209,
      "step": 1985
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.2691,
      "step": 1986
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.2247,
      "step": 1987
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.0309,
      "step": 1988
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.2279,
      "step": 1989
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.3365,
      "step": 1990
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0002,
      "loss": 1.1724,
      "step": 1991
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 0.9617,
      "step": 1992
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.0238,
      "step": 1993
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0002,
      "loss": 1.2358,
      "step": 1994
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.3989,
      "step": 1995
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 0.9649,
      "step": 1996
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.2347,
      "step": 1997
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.4647,
      "step": 1998
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.1593,
      "step": 1999
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 1.3194,
      "step": 2000
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.2417,
      "step": 2001
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.2345,
      "step": 2002
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.2375,
      "step": 2003
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.048,
      "step": 2004
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.4147,
      "step": 2005
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.2475,
      "step": 2006
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.3431,
      "step": 2007
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.1238,
      "step": 2008
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.4866,
      "step": 2009
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.314,
      "step": 2010
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.3465,
      "step": 2011
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.0749,
      "step": 2012
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 0.8763,
      "step": 2013
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 0.7786,
      "step": 2014
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.1226,
      "step": 2015
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0002,
      "loss": 0.9459,
      "step": 2016
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.0523,
      "step": 2017
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.1292,
      "step": 2018
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.8902,
      "step": 2019
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.2388,
      "step": 2020
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.2953,
      "step": 2021
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.2769,
      "step": 2022
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.2398,
      "step": 2023
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.1369,
      "step": 2024
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.9375,
      "step": 2025
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.1164,
      "step": 2026
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.108,
      "step": 2027
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.2266,
      "step": 2028
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 0.7522,
      "step": 2029
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.2076,
      "step": 2030
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.3121,
      "step": 2031
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.0789,
      "step": 2032
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.8857,
      "step": 2033
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.0211,
      "step": 2034
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.0218,
      "step": 2035
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.0853,
      "step": 2036
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.3998,
      "step": 2037
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.9306,
      "step": 2038
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 0.8593,
      "step": 2039
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.2347,
      "step": 2040
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0002,
      "loss": 0.8281,
      "step": 2041
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.8496,
      "step": 2042
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.462,
      "step": 2043
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.2952,
      "step": 2044
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.7446,
      "step": 2045
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1556,
      "step": 2046
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0521,
      "step": 2047
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0002,
      "loss": 1.1794,
      "step": 2048
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 0.9113,
      "step": 2049
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 1.1646,
      "step": 2050
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.0051,
      "step": 2051
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0002,
      "loss": 1.2129,
      "step": 2052
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.0744,
      "step": 2053
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.2061,
      "step": 2054
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.161,
      "step": 2055
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.3319,
      "step": 2056
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.2543,
      "step": 2057
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.3457,
      "step": 2058
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.1391,
      "step": 2059
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.9045,
      "step": 2060
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0002,
      "loss": 0.8726,
      "step": 2061
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.1639,
      "step": 2062
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.2898,
      "step": 2063
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.7583,
      "step": 2064
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.1708,
      "step": 2065
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.205,
      "step": 2066
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.9108,
      "step": 2067
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 0.816,
      "step": 2068
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0002,
      "loss": 0.9005,
      "step": 2069
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.5748,
      "step": 2070
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.0494,
      "step": 2071
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.1213,
      "step": 2072
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.012,
      "step": 2073
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 1.2301,
      "step": 2074
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.1232,
      "step": 2075
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.2625,
      "step": 2076
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.0971,
      "step": 2077
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 1.2696,
      "step": 2078
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.09375,
      "learning_rate": 0.0002,
      "loss": 1.0,
      "step": 2079
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.1769,
      "step": 2080
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.7663,
      "step": 2081
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.4159,
      "step": 2082
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0002,
      "loss": 1.3142,
      "step": 2083
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0002,
      "loss": 1.0245,
      "step": 2084
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.6793,
      "step": 2085
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.0665,
      "step": 2086
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 1.4604,
      "step": 2087
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.1666,
      "step": 2088
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.0477,
      "step": 2089
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.2859,
      "step": 2090
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.375,
      "step": 2091
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 0.9476,
      "step": 2092
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1878,
      "step": 2093
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0002,
      "loss": 0.8163,
      "step": 2094
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.2209,
      "step": 2095
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 0.5128,
      "step": 2096
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.7291,
      "step": 2097
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.2618,
      "step": 2098
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.3509,
      "step": 2099
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.1673,
      "step": 2100
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.8221,
      "step": 2101
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.0284,
      "step": 2102
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.4346,
      "step": 2103
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0002,
      "loss": 0.7872,
      "step": 2104
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.9928,
      "step": 2105
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 0.8616,
      "step": 2106
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0002,
      "loss": 1.2028,
      "step": 2107
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.9802,
      "step": 2108
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0002,
      "loss": 1.2273,
      "step": 2109
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.578125,
      "learning_rate": 0.0002,
      "loss": 1.236,
      "step": 2110
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.2456,
      "step": 2111
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.1735,
      "step": 2112
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.1537,
      "step": 2113
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.3422,
      "step": 2114
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0002,
      "loss": 0.9099,
      "step": 2115
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 0.953,
      "step": 2116
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.2581,
      "step": 2117
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 0.9797,
      "step": 2118
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.2255,
      "step": 2119
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.3285,
      "step": 2120
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.0747,
      "step": 2121
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.4512,
      "step": 2122
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.9622,
      "step": 2123
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 0.8184,
      "step": 2124
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.2881,
      "step": 2125
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.105,
      "step": 2126
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.1049,
      "step": 2127
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 0.4474,
      "step": 2128
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.2548,
      "step": 2129
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 0.5459,
      "step": 2130
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 0.9343,
      "step": 2131
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.0381,
      "step": 2132
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.8834,
      "step": 2133
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.4192,
      "step": 2134
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.9882,
      "step": 2135
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 0.9601,
      "step": 2136
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.2903,
      "step": 2137
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.2864,
      "step": 2138
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.155,
      "step": 2139
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.6143,
      "step": 2140
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.2061,
      "step": 2141
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 0.8001,
      "step": 2142
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.1963,
      "step": 2143
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.9026,
      "step": 2144
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 1.0864,
      "step": 2145
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.1813,
      "step": 2146
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.1471,
      "step": 2147
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.9745,
      "step": 2148
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.2754,
      "step": 2149
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.2625,
      "step": 2150
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.194,
      "step": 2151
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 0.886,
      "step": 2152
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.5804,
      "step": 2153
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.1428,
      "step": 2154
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.959,
      "step": 2155
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.9974,
      "step": 2156
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.1671,
      "step": 2157
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.5496,
      "step": 2158
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.5124,
      "step": 2159
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.306,
      "step": 2160
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.08056640625,
      "learning_rate": 0.0002,
      "loss": 0.3446,
      "step": 2161
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 0.7595,
      "step": 2162
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 0.9794,
      "step": 2163
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.0328,
      "step": 2164
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.0254,
      "step": 2165
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 0.7605,
      "step": 2166
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.0292,
      "step": 2167
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.9869,
      "step": 2168
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.1995,
      "step": 2169
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.047,
      "step": 2170
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.9918,
      "step": 2171
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.6036,
      "step": 2172
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 0.9724,
      "step": 2173
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.091,
      "step": 2174
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0453,
      "step": 2175
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.0149,
      "step": 2176
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.9523,
      "step": 2177
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 0.9713,
      "step": 2178
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.1213,
      "step": 2179
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.2364,
      "step": 2180
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 0.9065,
      "step": 2181
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.1193,
      "step": 2182
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 0.9857,
      "step": 2183
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 0.9898,
      "step": 2184
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.2199,
      "step": 2185
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.3202,
      "step": 2186
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.3007,
      "step": 2187
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.5254,
      "step": 2188
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.3141,
      "step": 2189
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.3151,
      "step": 2190
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.613,
      "step": 2191
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 1.1906,
      "step": 2192
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.0571,
      "step": 2193
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.1915,
      "step": 2194
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.2687,
      "step": 2195
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0002,
      "loss": 0.9162,
      "step": 2196
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.0926,
      "step": 2197
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.3116,
      "step": 2198
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0002,
      "loss": 1.4435,
      "step": 2199
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0002,
      "loss": 0.99,
      "step": 2200
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.4314,
      "step": 2201
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.2157,
      "step": 2202
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.1091,
      "step": 2203
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.2542,
      "step": 2204
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.4361,
      "step": 2205
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.511,
      "step": 2206
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.1677,
      "step": 2207
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 1.2599,
      "step": 2208
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.3513,
      "step": 2209
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 1.1621,
      "step": 2210
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0002,
      "loss": 0.7984,
      "step": 2211
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.07568359375,
      "learning_rate": 0.0002,
      "loss": 0.4905,
      "step": 2212
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 0.9904,
      "step": 2213
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.162,
      "step": 2214
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.152,
      "step": 2215
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0002,
      "loss": 1.0365,
      "step": 2216
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.1996,
      "step": 2217
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.3208,
      "step": 2218
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.9994,
      "step": 2219
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.1314,
      "step": 2220
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.082,
      "step": 2221
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 0.8128,
      "step": 2222
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.0756,
      "step": 2223
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.3516,
      "step": 2224
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 0.8938,
      "step": 2225
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.8165,
      "step": 2226
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 0.8942,
      "step": 2227
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.3562,
      "step": 2228
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.2637,
      "step": 2229
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 0.8121,
      "step": 2230
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.8707,
      "step": 2231
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.456,
      "step": 2232
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 0.8955,
      "step": 2233
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 0.9699,
      "step": 2234
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.3742,
      "step": 2235
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 0.7521,
      "step": 2236
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.9829,
      "step": 2237
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.1551,
      "step": 2238
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.0563,
      "step": 2239
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.2632,
      "step": 2240
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.3459,
      "step": 2241
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.1513,
      "step": 2242
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1839,
      "step": 2243
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 0.8779,
      "step": 2244
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0002,
      "loss": 0.9074,
      "step": 2245
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 0.865,
      "step": 2246
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.4629,
      "step": 2247
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 0.7659,
      "step": 2248
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0002,
      "loss": 0.7468,
      "step": 2249
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.5319,
      "step": 2250
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.0975,
      "step": 2251
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.6217,
      "step": 2252
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.3168,
      "step": 2253
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.765,
      "step": 2254
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.1008,
      "step": 2255
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.4452,
      "step": 2256
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 0.6449,
      "step": 2257
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 0.8177,
      "step": 2258
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.1726,
      "step": 2259
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.1805,
      "step": 2260
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.3016,
      "step": 2261
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.3838,
      "step": 2262
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.9652,
      "step": 2263
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.614,
      "step": 2264
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.4154,
      "step": 2265
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.0856,
      "step": 2266
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.2475,
      "step": 2267
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.3038,
      "step": 2268
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.9991,
      "step": 2269
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.0908,
      "step": 2270
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 0.8056,
      "step": 2271
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 0.9403,
      "step": 2272
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1305,
      "step": 2273
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 0.8102,
      "step": 2274
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 0.959,
      "step": 2275
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.0145,
      "step": 2276
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.7549,
      "step": 2277
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.0523,
      "step": 2278
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3309,
      "step": 2279
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 0.9318,
      "step": 2280
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.131,
      "step": 2281
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.8511,
      "step": 2282
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.9278,
      "step": 2283
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 1.4364,
      "step": 2284
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.2583,
      "step": 2285
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0002,
      "loss": 1.0145,
      "step": 2286
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.3347,
      "step": 2287
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.107,
      "step": 2288
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.0236,
      "step": 2289
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 0.4444,
      "step": 2290
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.2606,
      "step": 2291
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.8287,
      "step": 2292
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0002,
      "loss": 1.0687,
      "step": 2293
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 0.8991,
      "step": 2294
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2336,
      "step": 2295
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.7672,
      "step": 2296
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.3929,
      "step": 2297
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.168,
      "step": 2298
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 0.9737,
      "step": 2299
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0002,
      "loss": 1.1945,
      "step": 2300
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.1954,
      "step": 2301
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.334,
      "step": 2302
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 0.9912,
      "step": 2303
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.0169,
      "step": 2304
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0002,
      "loss": 0.8827,
      "step": 2305
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.2874,
      "step": 2306
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.1961,
      "step": 2307
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0002,
      "loss": 1.3617,
      "step": 2308
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.3845,
      "step": 2309
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.1323,
      "step": 2310
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 1.0067,
      "step": 2311
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.0095,
      "step": 2312
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.8707,
      "step": 2313
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 2.1368,
      "step": 2314
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 0.7482,
      "step": 2315
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.9813,
      "step": 2316
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0002,
      "loss": 1.3767,
      "step": 2317
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 0.4632,
      "step": 2318
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.6572,
      "step": 2319
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0002,
      "loss": 1.2366,
      "step": 2320
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.0314,
      "step": 2321
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 0.8611,
      "step": 2322
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.8615,
      "step": 2323
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.09375,
      "learning_rate": 0.0002,
      "loss": 0.9536,
      "step": 2324
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.4205,
      "step": 2325
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.2911,
      "step": 2326
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 0.5499,
      "step": 2327
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.0164,
      "step": 2328
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.5856,
      "step": 2329
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.0007,
      "step": 2330
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0002,
      "loss": 0.9506,
      "step": 2331
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.1785,
      "step": 2332
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.1868,
      "step": 2333
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 0.8918,
      "step": 2334
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.481,
      "step": 2335
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.9326,
      "step": 2336
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.0527,
      "step": 2337
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 0.7227,
      "step": 2338
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3864,
      "step": 2339
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.4831,
      "step": 2340
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 0.9119,
      "step": 2341
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.356,
      "step": 2342
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.7575,
      "step": 2343
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 1.2677,
      "step": 2344
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.0868,
      "step": 2345
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.1874,
      "step": 2346
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0002,
      "loss": 1.0038,
      "step": 2347
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.0544,
      "step": 2348
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 0.786,
      "step": 2349
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.5317,
      "step": 2350
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.128,
      "step": 2351
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0002,
      "loss": 0.3059,
      "step": 2352
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.5284,
      "step": 2353
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.3855,
      "step": 2354
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1589,
      "step": 2355
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.312,
      "step": 2356
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.2149,
      "step": 2357
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.1322,
      "step": 2358
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 0.909,
      "step": 2359
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.3328,
      "step": 2360
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.8028,
      "step": 2361
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0002,
      "loss": 1.2284,
      "step": 2362
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.2206,
      "step": 2363
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.3177,
      "step": 2364
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.0213,
      "step": 2365
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.5985,
      "step": 2366
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.8716,
      "step": 2367
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0002,
      "loss": 1.0358,
      "step": 2368
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.8624,
      "step": 2369
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.1592,
      "step": 2370
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.4494,
      "step": 2371
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.1886,
      "step": 2372
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 0.8554,
      "step": 2373
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 0.9017,
      "step": 2374
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 0.896,
      "step": 2375
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.3212,
      "step": 2376
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3266,
      "step": 2377
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.1765,
      "step": 2378
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.3731,
      "step": 2379
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.0622,
      "step": 2380
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 1.15,
      "step": 2381
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.2772,
      "step": 2382
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 0.9355,
      "step": 2383
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.3489,
      "step": 2384
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.3897,
      "step": 2385
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.246,
      "step": 2386
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0002,
      "loss": 1.4064,
      "step": 2387
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.0928,
      "step": 2388
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 0.9512,
      "step": 2389
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 0.9689,
      "step": 2390
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 0.8884,
      "step": 2391
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.4638,
      "step": 2392
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2028,
      "step": 2393
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.1639,
      "step": 2394
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.0133,
      "step": 2395
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.8859,
      "step": 2396
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.9756,
      "step": 2397
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.8405,
      "step": 2398
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 0.405,
      "step": 2399
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.2797,
      "step": 2400
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.0019,
      "step": 2401
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.0531,
      "step": 2402
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.6497,
      "step": 2403
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.9781,
      "step": 2404
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.2021,
      "step": 2405
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.1265,
      "step": 2406
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.2459,
      "step": 2407
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 0.9484,
      "step": 2408
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.9191,
      "step": 2409
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.9884,
      "step": 2410
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0002,
      "loss": 0.7564,
      "step": 2411
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.2493,
      "step": 2412
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.5137,
      "step": 2413
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.1527,
      "step": 2414
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.8087,
      "step": 2415
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.2682,
      "step": 2416
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.222,
      "step": 2417
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.354,
      "step": 2418
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.2355,
      "step": 2419
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.0559,
      "step": 2420
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0002,
      "loss": 1.0025,
      "step": 2421
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.9417,
      "step": 2422
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.4146,
      "step": 2423
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3617,
      "step": 2424
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.3058,
      "step": 2425
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0325,
      "step": 2426
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 1.3337,
      "step": 2427
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0002,
      "loss": 1.3069,
      "step": 2428
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 0.878,
      "step": 2429
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 0.8913,
      "step": 2430
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.0752,
      "step": 2431
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0002,
      "loss": 0.7291,
      "step": 2432
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.4112,
      "step": 2433
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.0954,
      "step": 2434
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 0.8537,
      "step": 2435
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.9799,
      "step": 2436
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.1288,
      "step": 2437
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.4974,
      "step": 2438
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.9935,
      "step": 2439
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 0.9751,
      "step": 2440
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.4436,
      "step": 2441
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 0.9216,
      "step": 2442
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 0.9895,
      "step": 2443
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0002,
      "loss": 1.2605,
      "step": 2444
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.0785,
      "step": 2445
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.3392,
      "step": 2446
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.2307,
      "step": 2447
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.9377,
      "step": 2448
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0002,
      "loss": 0.5054,
      "step": 2449
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.8455,
      "step": 2450
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 0.4683,
      "step": 2451
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.2053,
      "step": 2452
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.4321,
      "step": 2453
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.39,
      "step": 2454
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.0097,
      "step": 2455
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.3112,
      "step": 2456
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.1354,
      "step": 2457
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.1065,
      "step": 2458
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.1071,
      "step": 2459
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.1006,
      "step": 2460
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.0341,
      "step": 2461
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.2091,
      "step": 2462
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.2798,
      "step": 2463
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.3291,
      "step": 2464
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.9463,
      "step": 2465
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.2979,
      "step": 2466
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0002,
      "loss": 1.0474,
      "step": 2467
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.166,
      "step": 2468
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.1429,
      "step": 2469
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.9476,
      "step": 2470
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 1.0329,
      "step": 2471
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.1993,
      "step": 2472
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.6433,
      "step": 2473
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 0.9401,
      "step": 2474
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.1845,
      "step": 2475
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.3301,
      "step": 2476
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.3278,
      "step": 2477
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 0.7769,
      "step": 2478
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.5077,
      "step": 2479
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.3144,
      "step": 2480
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.1871,
      "step": 2481
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.3381,
      "step": 2482
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.3287,
      "step": 2483
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.1154,
      "step": 2484
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 1.1035,
      "step": 2485
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.0545,
      "step": 2486
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.1015,
      "step": 2487
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.384,
      "step": 2488
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.1656,
      "step": 2489
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.7895,
      "step": 2490
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.3124,
      "step": 2491
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.2853,
      "step": 2492
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.0031,
      "step": 2493
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.2885,
      "step": 2494
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0002,
      "loss": 1.149,
      "step": 2495
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0002,
      "loss": 1.0391,
      "step": 2496
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.0245,
      "step": 2497
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.3311,
      "step": 2498
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.1909,
      "step": 2499
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 1.0822,
      "step": 2500
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.4502,
      "step": 2501
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.0725,
      "step": 2502
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.3099,
      "step": 2503
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.7359,
      "step": 2504
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.9909,
      "step": 2505
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.4322,
      "step": 2506
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.5188,
      "step": 2507
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.1897,
      "step": 2508
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 0.9645,
      "step": 2509
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 0.9674,
      "step": 2510
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 0.8995,
      "step": 2511
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.7271,
      "step": 2512
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.1743,
      "step": 2513
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.1537,
      "step": 2514
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 1.2778,
      "step": 2515
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.7892,
      "step": 2516
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.5148,
      "step": 2517
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.3381,
      "step": 2518
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.0415,
      "step": 2519
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.9328,
      "step": 2520
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.3009,
      "step": 2521
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.9708,
      "step": 2522
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.224,
      "step": 2523
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.175,
      "step": 2524
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 0.8073,
      "step": 2525
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.2145,
      "step": 2526
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.3606,
      "step": 2527
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.5531,
      "step": 2528
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 0.9227,
      "step": 2529
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.7981,
      "step": 2530
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.0962,
      "step": 2531
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.0923,
      "step": 2532
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 0.9119,
      "step": 2533
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.3202,
      "step": 2534
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.8729,
      "step": 2535
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.3547,
      "step": 2536
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.2341,
      "step": 2537
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.0067,
      "step": 2538
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 0.8103,
      "step": 2539
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.3374,
      "step": 2540
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 0.3904,
      "step": 2541
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.172,
      "step": 2542
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 0.9386,
      "step": 2543
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.193,
      "step": 2544
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.0668,
      "step": 2545
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0002,
      "loss": 1.2911,
      "step": 2546
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.0577,
      "step": 2547
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0737,
      "step": 2548
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.2876,
      "step": 2549
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.3191,
      "step": 2550
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.7385,
      "step": 2551
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.0256,
      "step": 2552
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.3411,
      "step": 2553
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.7993,
      "step": 2554
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.097,
      "step": 2555
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0105,
      "step": 2556
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.7635,
      "step": 2557
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.0074,
      "step": 2558
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0002,
      "loss": 1.1643,
      "step": 2559
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.1096,
      "step": 2560
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.5949,
      "step": 2561
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.1069,
      "step": 2562
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.0725,
      "step": 2563
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0002,
      "loss": 0.7884,
      "step": 2564
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.1898,
      "step": 2565
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.1554,
      "step": 2566
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 0.7725,
      "step": 2567
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.917,
      "step": 2568
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.0919,
      "step": 2569
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.0959,
      "step": 2570
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 0.8216,
      "step": 2571
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.4125,
      "step": 2572
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.3102,
      "step": 2573
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 0.854,
      "step": 2574
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.0425,
      "step": 2575
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 0.7784,
      "step": 2576
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.2668,
      "step": 2577
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0002,
      "loss": 0.9709,
      "step": 2578
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.3028,
      "step": 2579
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 0.6176,
      "step": 2580
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.1205,
      "step": 2581
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.0091,
      "step": 2582
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.257,
      "step": 2583
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 1.237,
      "step": 2584
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.254,
      "step": 2585
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.1587,
      "step": 2586
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.2604,
      "step": 2587
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.2888,
      "step": 2588
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.2163,
      "step": 2589
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2372,
      "step": 2590
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0017,
      "step": 2591
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.0874,
      "step": 2592
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.606,
      "step": 2593
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.0971,
      "step": 2594
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.1344,
      "step": 2595
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.0168,
      "step": 2596
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.2611,
      "step": 2597
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.969,
      "step": 2598
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.2122,
      "step": 2599
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.1468,
      "step": 2600
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.9889,
      "step": 2601
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.8113,
      "step": 2602
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.0212,
      "step": 2603
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.1235,
      "step": 2604
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.5616,
      "step": 2605
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.3706,
      "step": 2606
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.1453,
      "step": 2607
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.9483,
      "step": 2608
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.1235,
      "step": 2609
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2318,
      "step": 2610
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.233,
      "step": 2611
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0002,
      "loss": 0.9342,
      "step": 2612
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 1.1697,
      "step": 2613
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.1011,
      "step": 2614
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.1359,
      "step": 2615
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.1541,
      "step": 2616
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 1.2753,
      "step": 2617
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.0281,
      "step": 2618
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.0402,
      "step": 2619
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.2698,
      "step": 2620
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.9166,
      "step": 2621
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0002,
      "loss": 0.6829,
      "step": 2622
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 1.0419,
      "step": 2623
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 0.8339,
      "step": 2624
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.0675,
      "step": 2625
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.1823,
      "step": 2626
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 0.8622,
      "step": 2627
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.96,
      "step": 2628
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.43,
      "step": 2629
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.8405,
      "step": 2630
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.1631,
      "step": 2631
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.6978,
      "step": 2632
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.9867,
      "step": 2633
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.0625,
      "step": 2634
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.5297,
      "step": 2635
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0002,
      "loss": 1.3967,
      "step": 2636
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.4757,
      "step": 2637
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0002,
      "loss": 1.3929,
      "step": 2638
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.4862,
      "step": 2639
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.9684,
      "step": 2640
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.245,
      "step": 2641
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1932,
      "step": 2642
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.1206,
      "step": 2643
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.9033,
      "step": 2644
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1101,
      "step": 2645
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 0.6802,
      "step": 2646
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.4369,
      "step": 2647
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.4431,
      "step": 2648
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.0595,
      "step": 2649
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 0.9556,
      "step": 2650
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.1532,
      "step": 2651
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.9591,
      "step": 2652
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.5098,
      "step": 2653
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.2338,
      "step": 2654
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 1.1277,
      "step": 2655
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.3344,
      "step": 2656
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 0.571,
      "step": 2657
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.5765,
      "step": 2658
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.2872,
      "step": 2659
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1681,
      "step": 2660
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 0.7177,
      "step": 2661
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.3751,
      "step": 2662
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.1886,
      "step": 2663
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 0.7412,
      "step": 2664
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.3566,
      "step": 2665
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.1787,
      "step": 2666
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.2547,
      "step": 2667
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2651,
      "step": 2668
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 0.8505,
      "step": 2669
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0002,
      "loss": 0.9748,
      "step": 2670
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.1924,
      "step": 2671
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.9181,
      "step": 2672
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.0767,
      "step": 2673
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.3854,
      "step": 2674
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.9131,
      "step": 2675
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.2822,
      "step": 2676
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0002,
      "loss": 0.9539,
      "step": 2677
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 0.8069,
      "step": 2678
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.0987,
      "step": 2679
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.8794,
      "step": 2680
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.7672,
      "step": 2681
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0002,
      "loss": 0.7176,
      "step": 2682
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.9774,
      "step": 2683
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.2927,
      "step": 2684
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.3223,
      "step": 2685
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.9109,
      "step": 2686
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 0.9237,
      "step": 2687
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 0.8354,
      "step": 2688
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 0.9576,
      "step": 2689
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.0927,
      "step": 2690
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3833,
      "step": 2691
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.0181,
      "step": 2692
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0002,
      "loss": 1.0786,
      "step": 2693
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.4684,
      "step": 2694
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.3825,
      "step": 2695
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.8445,
      "step": 2696
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.7179,
      "step": 2697
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.2534,
      "step": 2698
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 0.7813,
      "step": 2699
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0217,
      "step": 2700
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.2402,
      "step": 2701
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.6788,
      "step": 2702
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0264,
      "step": 2703
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.1238,
      "step": 2704
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.1769,
      "step": 2705
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.2524,
      "step": 2706
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0002,
      "loss": 0.7341,
      "step": 2707
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0002,
      "loss": 0.9766,
      "step": 2708
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.09326171875,
      "learning_rate": 0.0002,
      "loss": 1.0967,
      "step": 2709
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.2567,
      "step": 2710
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.0028,
      "step": 2711
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 1.3006,
      "step": 2712
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 1.0519,
      "step": 2713
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.7835,
      "step": 2714
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.5651,
      "step": 2715
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.2688,
      "step": 2716
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.1762,
      "step": 2717
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0002,
      "loss": 0.9326,
      "step": 2718
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.0143,
      "step": 2719
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.9812,
      "step": 2720
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.0365,
      "step": 2721
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.3412,
      "step": 2722
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.1487,
      "step": 2723
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.458,
      "step": 2724
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.0738,
      "step": 2725
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0002,
      "loss": 0.6052,
      "step": 2726
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.397,
      "step": 2727
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.1743,
      "step": 2728
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.0474,
      "step": 2729
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.0728,
      "step": 2730
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.8184,
      "step": 2731
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1823,
      "step": 2732
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0002,
      "loss": 1.5171,
      "step": 2733
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.0962,
      "step": 2734
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.0052,
      "step": 2735
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.1952,
      "step": 2736
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.1984,
      "step": 2737
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.0236,
      "step": 2738
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.1167,
      "step": 2739
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 0.795,
      "step": 2740
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.2198,
      "step": 2741
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.0668,
      "step": 2742
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.0108,
      "step": 2743
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.1906,
      "step": 2744
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.1097,
      "step": 2745
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.792,
      "step": 2746
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.4875,
      "step": 2747
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.4625,
      "step": 2748
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.3358,
      "step": 2749
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.1837,
      "step": 2750
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0002,
      "loss": 0.5499,
      "step": 2751
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 0.7353,
      "step": 2752
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.1702,
      "step": 2753
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.1071,
      "step": 2754
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.0902,
      "step": 2755
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.178,
      "step": 2756
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.1171,
      "step": 2757
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 1.027,
      "step": 2758
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 0.9452,
      "step": 2759
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.987,
      "step": 2760
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.2216,
      "step": 2761
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.8459,
      "step": 2762
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.7331,
      "step": 2763
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.5225,
      "step": 2764
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.1283,
      "step": 2765
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0002,
      "loss": 0.8667,
      "step": 2766
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.066,
      "step": 2767
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1033,
      "step": 2768
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2595,
      "step": 2769
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.4946,
      "step": 2770
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0002,
      "loss": 0.649,
      "step": 2771
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0002,
      "loss": 1.1522,
      "step": 2772
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.3552,
      "step": 2773
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.3315,
      "step": 2774
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.1311,
      "step": 2775
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0002,
      "loss": 0.7148,
      "step": 2776
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.1244,
      "step": 2777
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 0.7198,
      "step": 2778
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.9137,
      "step": 2779
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.0901,
      "step": 2780
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 0.9926,
      "step": 2781
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1079,
      "step": 2782
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.9994,
      "step": 2783
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.2724,
      "step": 2784
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0002,
      "loss": 1.6693,
      "step": 2785
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.1139,
      "step": 2786
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0002,
      "loss": 0.8952,
      "step": 2787
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 0.7897,
      "step": 2788
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.882,
      "step": 2789
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.5455,
      "step": 2790
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.9019,
      "step": 2791
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.0816,
      "step": 2792
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 0.7715,
      "step": 2793
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.3147,
      "step": 2794
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.2184,
      "step": 2795
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.0011,
      "step": 2796
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3057,
      "step": 2797
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.6858,
      "step": 2798
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.2883,
      "step": 2799
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.0624,
      "step": 2800
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.1591,
      "step": 2801
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.0684,
      "step": 2802
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.296875,
      "learning_rate": 0.0002,
      "loss": 1.0644,
      "step": 2803
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.0321,
      "step": 2804
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.4167,
      "step": 2805
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3335,
      "step": 2806
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 1.3026,
      "step": 2807
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.3811,
      "step": 2808
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.4075,
      "step": 2809
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.9617,
      "step": 2810
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.3305,
      "step": 2811
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 0.9442,
      "step": 2812
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.108,
      "step": 2813
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.9115,
      "step": 2814
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0002,
      "loss": 0.6413,
      "step": 2815
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.5295,
      "step": 2816
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.2683,
      "step": 2817
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.1898,
      "step": 2818
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.7558,
      "step": 2819
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.6381,
      "step": 2820
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.086,
      "step": 2821
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.09375,
      "learning_rate": 0.0002,
      "loss": 0.5832,
      "step": 2822
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.9717,
      "step": 2823
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 0.8358,
      "step": 2824
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.3208,
      "step": 2825
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.1036,
      "step": 2826
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.1181,
      "step": 2827
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.449,
      "step": 2828
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.1454,
      "step": 2829
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.8721,
      "step": 2830
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 0.9275,
      "step": 2831
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.2153,
      "step": 2832
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0002,
      "loss": 0.6402,
      "step": 2833
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.1556,
      "step": 2834
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.1991,
      "step": 2835
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.1611,
      "step": 2836
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.3209,
      "step": 2837
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 0.9367,
      "step": 2838
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.2101,
      "step": 2839
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.7584,
      "step": 2840
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.356,
      "step": 2841
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.0273,
      "step": 2842
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.2221,
      "step": 2843
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2015,
      "step": 2844
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.0048,
      "step": 2845
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.3895,
      "step": 2846
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2459,
      "step": 2847
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.1185,
      "step": 2848
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.1356,
      "step": 2849
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.0912,
      "step": 2850
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0002,
      "loss": 0.627,
      "step": 2851
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.9399,
      "step": 2852
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.0227,
      "step": 2853
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 0.9948,
      "step": 2854
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.8326,
      "step": 2855
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.1127,
      "step": 2856
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.2506,
      "step": 2857
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.2258,
      "step": 2858
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0002,
      "loss": 1.0386,
      "step": 2859
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.9541,
      "step": 2860
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 0.6869,
      "step": 2861
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.0833,
      "step": 2862
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 0.9316,
      "step": 2863
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.1007,
      "step": 2864
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.1384,
      "step": 2865
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.157,
      "step": 2866
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.1597,
      "step": 2867
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.2059,
      "step": 2868
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.2682,
      "step": 2869
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 0.9738,
      "step": 2870
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1854,
      "step": 2871
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.1543,
      "step": 2872
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.2533,
      "step": 2873
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 0.9271,
      "step": 2874
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.2512,
      "step": 2875
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 0.8328,
      "step": 2876
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.2897,
      "step": 2877
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 1.1637,
      "step": 2878
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.196,
      "step": 2879
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.3012,
      "step": 2880
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.8655,
      "step": 2881
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 0.4344,
      "step": 2882
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.1994,
      "step": 2883
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.1336,
      "step": 2884
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 0.911,
      "step": 2885
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.2533,
      "step": 2886
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.4109,
      "step": 2887
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.2727,
      "step": 2888
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 1.2887,
      "step": 2889
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.1343,
      "step": 2890
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.1588,
      "step": 2891
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.1934,
      "step": 2892
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1406,
      "step": 2893
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.1549,
      "step": 2894
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.9925,
      "step": 2895
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0866,
      "step": 2896
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.0589,
      "step": 2897
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1552,
      "step": 2898
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.881,
      "step": 2899
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0002,
      "loss": 0.8599,
      "step": 2900
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.52,
      "step": 2901
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.1601,
      "step": 2902
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 0.9996,
      "step": 2903
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.2488,
      "step": 2904
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.8579,
      "step": 2905
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.417,
      "step": 2906
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.4009,
      "step": 2907
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.632,
      "step": 2908
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.3653,
      "step": 2909
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 0.9022,
      "step": 2910
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.0897,
      "step": 2911
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.9711,
      "step": 2912
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.1914,
      "step": 2913
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.0844,
      "step": 2914
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 1.4159,
      "step": 2915
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.4094,
      "step": 2916
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 1.1977,
      "step": 2917
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.1644,
      "step": 2918
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.3809,
      "step": 2919
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.3744,
      "step": 2920
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.295,
      "step": 2921
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.0893,
      "step": 2922
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.0453,
      "step": 2923
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.6319,
      "step": 2924
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.4018,
      "step": 2925
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0002,
      "loss": 0.7908,
      "step": 2926
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.2735,
      "step": 2927
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 0.8674,
      "step": 2928
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.3056,
      "step": 2929
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 1.0383,
      "step": 2930
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 0.8468,
      "step": 2931
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.3551,
      "step": 2932
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.4103,
      "step": 2933
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 1.0044,
      "step": 2934
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.8639,
      "step": 2935
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.2878,
      "step": 2936
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.5595,
      "step": 2937
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.2972,
      "step": 2938
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.4242,
      "step": 2939
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1844,
      "step": 2940
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.1366,
      "step": 2941
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.1576,
      "step": 2942
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 0.9806,
      "step": 2943
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.2474,
      "step": 2944
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 0.9678,
      "step": 2945
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 0.9472,
      "step": 2946
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 1.5849,
      "step": 2947
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3544,
      "step": 2948
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.3695,
      "step": 2949
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1447,
      "step": 2950
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.8567,
      "step": 2951
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0967,
      "step": 2952
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.4616,
      "step": 2953
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.0308,
      "step": 2954
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.2941,
      "step": 2955
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 0.7281,
      "step": 2956
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.0464,
      "step": 2957
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.0795,
      "step": 2958
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.0907,
      "step": 2959
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 0.556,
      "step": 2960
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 0.9239,
      "step": 2961
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.0143,
      "step": 2962
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 0.9279,
      "step": 2963
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.147,
      "step": 2964
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.1619,
      "step": 2965
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.9439,
      "step": 2966
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.1765,
      "step": 2967
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.1009,
      "step": 2968
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.1878,
      "step": 2969
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.3212,
      "step": 2970
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.1616,
      "step": 2971
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0002,
      "loss": 0.6099,
      "step": 2972
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.0249,
      "step": 2973
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.1113,
      "step": 2974
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.2934,
      "step": 2975
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.1789,
      "step": 2976
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.2602,
      "step": 2977
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.958,
      "step": 2978
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.3986,
      "step": 2979
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.241,
      "step": 2980
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.9117,
      "step": 2981
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 0.5187,
      "step": 2982
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.0805,
      "step": 2983
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.1631,
      "step": 2984
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.1238,
      "step": 2985
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.237,
      "step": 2986
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.8486,
      "step": 2987
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.0443,
      "step": 2988
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.364,
      "step": 2989
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.7477,
      "step": 2990
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.217,
      "step": 2991
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.0814,
      "step": 2992
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.2239,
      "step": 2993
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.031,
      "step": 2994
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 0.9516,
      "step": 2995
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.0913,
      "step": 2996
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.9688,
      "step": 2997
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.1242,
      "step": 2998
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0002,
      "loss": 0.8613,
      "step": 2999
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 0.8923,
      "step": 3000
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.0089,
      "step": 3001
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.9011,
      "step": 3002
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.1511,
      "step": 3003
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.2086,
      "step": 3004
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9296875,
      "learning_rate": 0.0002,
      "loss": 1.9083,
      "step": 3005
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0002,
      "loss": 0.5015,
      "step": 3006
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.0844,
      "step": 3007
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 0.7796,
      "step": 3008
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.1655,
      "step": 3009
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.1798,
      "step": 3010
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 0.96,
      "step": 3011
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.7715,
      "step": 3012
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.1502,
      "step": 3013
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.1795,
      "step": 3014
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 0.5222,
      "step": 3015
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3573,
      "step": 3016
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 1.5214,
      "step": 3017
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.7898,
      "step": 3018
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0002,
      "loss": 1.1444,
      "step": 3019
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0807,
      "step": 3020
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.0594,
      "step": 3021
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.0922,
      "step": 3022
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 0.8791,
      "step": 3023
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.2639,
      "step": 3024
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.5256,
      "step": 3025
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 0.9717,
      "step": 3026
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.3318,
      "step": 3027
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 0.9227,
      "step": 3028
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.9762,
      "step": 3029
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.9129,
      "step": 3030
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.7619,
      "step": 3031
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 0.9748,
      "step": 3032
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0943,
      "step": 3033
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.2611,
      "step": 3034
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.1635,
      "step": 3035
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.1769,
      "step": 3036
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.9672,
      "step": 3037
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.2256,
      "step": 3038
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.2096,
      "step": 3039
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.3792,
      "step": 3040
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 0.867,
      "step": 3041
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.1182,
      "step": 3042
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.4644,
      "step": 3043
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0002,
      "loss": 0.7859,
      "step": 3044
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.3537,
      "step": 3045
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.07958984375,
      "learning_rate": 0.0002,
      "loss": 0.751,
      "step": 3046
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.7028,
      "step": 3047
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.1418,
      "step": 3048
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 0.8939,
      "step": 3049
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.6029,
      "step": 3050
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.1075,
      "step": 3051
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.5702,
      "step": 3052
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 0.9021,
      "step": 3053
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1692,
      "step": 3054
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.3066,
      "step": 3055
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.2326,
      "step": 3056
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.1115,
      "step": 3057
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.8744,
      "step": 3058
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.3648,
      "step": 3059
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0002,
      "loss": 0.7755,
      "step": 3060
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.1823,
      "step": 3061
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.062,
      "step": 3062
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0716,
      "step": 3063
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.3646,
      "step": 3064
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 1.0959,
      "step": 3065
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 0.765,
      "step": 3066
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.1408,
      "step": 3067
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.1905,
      "step": 3068
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.9997,
      "step": 3069
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 0.702,
      "step": 3070
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.4815,
      "step": 3071
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.5182,
      "step": 3072
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 0.9887,
      "step": 3073
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.0974,
      "step": 3074
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 0.8432,
      "step": 3075
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.6366,
      "step": 3076
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.1933,
      "step": 3077
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.2126,
      "step": 3078
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 0.8688,
      "step": 3079
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.0571,
      "step": 3080
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 0.8691,
      "step": 3081
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.3552,
      "step": 3082
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.2643,
      "step": 3083
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 0.7762,
      "step": 3084
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 0.9804,
      "step": 3085
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.1287,
      "step": 3086
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.1803,
      "step": 3087
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0474,
      "step": 3088
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.262,
      "step": 3089
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.214,
      "step": 3090
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.1055,
      "step": 3091
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 0.8453,
      "step": 3092
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.1448,
      "step": 3093
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.8816,
      "step": 3094
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.9396,
      "step": 3095
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0002,
      "loss": 0.9788,
      "step": 3096
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.3052,
      "step": 3097
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0038,
      "step": 3098
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.2787,
      "step": 3099
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.35,
      "step": 3100
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.0094,
      "step": 3101
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.9328,
      "step": 3102
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 0.7552,
      "step": 3103
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.9847,
      "step": 3104
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.2,
      "step": 3105
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.2023,
      "step": 3106
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.0671,
      "step": 3107
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 1.258,
      "step": 3108
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 0.8346,
      "step": 3109
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.058,
      "step": 3110
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.1393,
      "step": 3111
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.1383,
      "step": 3112
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 0.999,
      "step": 3113
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 1.1659,
      "step": 3114
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.0657,
      "step": 3115
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.9749,
      "step": 3116
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1015,
      "step": 3117
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.316,
      "step": 3118
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.7737,
      "step": 3119
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.5016,
      "step": 3120
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 0.9401,
      "step": 3121
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.3513,
      "step": 3122
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.0574,
      "step": 3123
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.9304,
      "step": 3124
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.4088,
      "step": 3125
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.9178,
      "step": 3126
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0002,
      "loss": 0.7383,
      "step": 3127
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.0783,
      "step": 3128
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.3439,
      "step": 3129
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.9594,
      "step": 3130
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.7671,
      "step": 3131
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.1792,
      "step": 3132
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 0.9854,
      "step": 3133
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.5131,
      "step": 3134
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.8514,
      "step": 3135
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 0.7714,
      "step": 3136
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.1448,
      "step": 3137
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.0246,
      "step": 3138
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.1262,
      "step": 3139
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.1705,
      "step": 3140
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.0675,
      "step": 3141
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.2565,
      "step": 3142
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.9923,
      "step": 3143
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0558,
      "step": 3144
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.0021,
      "step": 3145
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.0024,
      "step": 3146
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 1.0119,
      "step": 3147
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.0004,
      "step": 3148
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 0.8371,
      "step": 3149
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.8333,
      "step": 3150
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.8853,
      "step": 3151
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.0835,
      "step": 3152
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 0.9191,
      "step": 3153
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.4111,
      "step": 3154
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 0.9372,
      "step": 3155
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.3387,
      "step": 3156
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.0188,
      "step": 3157
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.6252,
      "step": 3158
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.8666,
      "step": 3159
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.4944,
      "step": 3160
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0002,
      "loss": 0.5409,
      "step": 3161
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.2639,
      "step": 3162
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.259,
      "step": 3163
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.7725,
      "step": 3164
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.4677,
      "step": 3165
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.1694,
      "step": 3166
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.3683,
      "step": 3167
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2967,
      "step": 3168
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.103,
      "step": 3169
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.5551,
      "step": 3170
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 0.9355,
      "step": 3171
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 1.1298,
      "step": 3172
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.148,
      "step": 3173
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.327,
      "step": 3174
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.6449,
      "step": 3175
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 0.8403,
      "step": 3176
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.2889,
      "step": 3177
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.2585,
      "step": 3178
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.1198,
      "step": 3179
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.2907,
      "step": 3180
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.2988,
      "step": 3181
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0945,
      "step": 3182
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0002,
      "loss": 0.8127,
      "step": 3183
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 1.0525,
      "step": 3184
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.9319,
      "step": 3185
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.1171,
      "step": 3186
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.0613,
      "step": 3187
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 0.9826,
      "step": 3188
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.3149,
      "step": 3189
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2471,
      "step": 3190
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.1483,
      "step": 3191
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0583,
      "step": 3192
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.2099,
      "step": 3193
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.0302,
      "step": 3194
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.0048,
      "step": 3195
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.3973,
      "step": 3196
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.4016,
      "step": 3197
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0002,
      "loss": 0.7489,
      "step": 3198
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.8873,
      "step": 3199
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.2332,
      "step": 3200
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0002,
      "loss": 1.1404,
      "step": 3201
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.3197,
      "step": 3202
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.1822,
      "step": 3203
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 0.7048,
      "step": 3204
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.846,
      "step": 3205
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.2878,
      "step": 3206
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.9369,
      "step": 3207
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.9557,
      "step": 3208
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.1447,
      "step": 3209
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0002,
      "loss": 0.7984,
      "step": 3210
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.195,
      "step": 3211
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.2747,
      "step": 3212
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.261,
      "step": 3213
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 1.145,
      "step": 3214
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.0356,
      "step": 3215
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 0.9318,
      "step": 3216
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0002,
      "loss": 0.504,
      "step": 3217
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.4162,
      "step": 3218
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.0598,
      "step": 3219
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.0685,
      "step": 3220
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.3948,
      "step": 3221
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.3644,
      "step": 3222
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.8918,
      "step": 3223
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.4108,
      "step": 3224
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.3375,
      "step": 3225
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.2386,
      "step": 3226
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.2896,
      "step": 3227
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2353,
      "step": 3228
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 0.8066,
      "step": 3229
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.9859,
      "step": 3230
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.3079,
      "step": 3231
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0002,
      "loss": 0.9484,
      "step": 3232
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.1974,
      "step": 3233
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.2993,
      "step": 3234
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0226,
      "step": 3235
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.1336,
      "step": 3236
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0002,
      "loss": 1.2522,
      "step": 3237
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 1.3276,
      "step": 3238
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.052,
      "step": 3239
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.712,
      "step": 3240
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.4447,
      "step": 3241
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.3916,
      "step": 3242
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.2118,
      "step": 3243
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 0.4628,
      "step": 3244
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.2601,
      "step": 3245
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.9842,
      "step": 3246
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.5018,
      "step": 3247
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.8356,
      "step": 3248
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.6337,
      "step": 3249
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.8074,
      "step": 3250
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 0.8189,
      "step": 3251
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0002,
      "loss": 0.5548,
      "step": 3252
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0775,
      "step": 3253
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.199,
      "step": 3254
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.5488,
      "step": 3255
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.182,
      "step": 3256
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 0.9645,
      "step": 3257
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.4277,
      "step": 3258
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 0.8922,
      "step": 3259
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.2949,
      "step": 3260
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1547,
      "step": 3261
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 0.9466,
      "step": 3262
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.2534,
      "step": 3263
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.2555,
      "step": 3264
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.285,
      "step": 3265
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.3476,
      "step": 3266
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 0.6774,
      "step": 3267
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.1168,
      "step": 3268
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.3636,
      "step": 3269
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1345,
      "step": 3270
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.3296,
      "step": 3271
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.1195,
      "step": 3272
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 0.8083,
      "step": 3273
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.4553,
      "step": 3274
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.3336,
      "step": 3275
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.0017,
      "step": 3276
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 0.9761,
      "step": 3277
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 0.632,
      "step": 3278
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.0187,
      "step": 3279
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.272,
      "step": 3280
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.2586,
      "step": 3281
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.4087,
      "step": 3282
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2331,
      "step": 3283
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 1.0967,
      "step": 3284
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 0.8463,
      "step": 3285
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.0492,
      "step": 3286
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.3272,
      "step": 3287
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.8928,
      "step": 3288
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.8709,
      "step": 3289
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.1532,
      "step": 3290
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.0543,
      "step": 3291
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.1824,
      "step": 3292
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.9846,
      "step": 3293
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.995,
      "step": 3294
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.025,
      "step": 3295
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.0665,
      "step": 3296
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.0809,
      "step": 3297
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.1615,
      "step": 3298
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.0015,
      "step": 3299
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.5596,
      "step": 3300
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.4224,
      "step": 3301
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.0724,
      "step": 3302
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 1.1105,
      "step": 3303
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.1827,
      "step": 3304
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2063,
      "step": 3305
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.2625,
      "step": 3306
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.9984,
      "step": 3307
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.7637,
      "step": 3308
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.0948,
      "step": 3309
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.2889,
      "step": 3310
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.2858,
      "step": 3311
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.2207,
      "step": 3312
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0555,
      "step": 3313
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 1.2818,
      "step": 3314
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.4374,
      "step": 3315
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.3751,
      "step": 3316
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.9108,
      "step": 3317
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 0.6334,
      "step": 3318
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 0.8837,
      "step": 3319
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.9653,
      "step": 3320
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 0.9363,
      "step": 3321
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.3558,
      "step": 3322
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.2639,
      "step": 3323
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.2094,
      "step": 3324
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 0.5433,
      "step": 3325
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.2747,
      "step": 3326
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 0.7246,
      "step": 3327
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.1427,
      "step": 3328
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.9985,
      "step": 3329
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.1096,
      "step": 3330
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.6075,
      "step": 3331
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.7505,
      "step": 3332
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 1.2562,
      "step": 3333
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.1139,
      "step": 3334
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 1.0346,
      "step": 3335
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.0825,
      "step": 3336
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.0747,
      "step": 3337
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 0.8273,
      "step": 3338
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 0.8449,
      "step": 3339
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 0.9421,
      "step": 3340
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 0.8508,
      "step": 3341
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.287,
      "step": 3342
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002,
      "loss": 1.1773,
      "step": 3343
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.1406,
      "step": 3344
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2316,
      "step": 3345
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.163,
      "step": 3346
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.1163,
      "step": 3347
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.957,
      "step": 3348
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.2918,
      "step": 3349
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 0.7266,
      "step": 3350
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.6079,
      "step": 3351
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.133,
      "step": 3352
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.0485,
      "step": 3353
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.3049,
      "step": 3354
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.181,
      "step": 3355
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.2652,
      "step": 3356
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.6005,
      "step": 3357
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 0.9209,
      "step": 3358
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.2526,
      "step": 3359
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.0239,
      "step": 3360
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.8159,
      "step": 3361
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.2056,
      "step": 3362
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.1081,
      "step": 3363
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.104,
      "step": 3364
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.1223,
      "step": 3365
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.4361,
      "step": 3366
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 0.9752,
      "step": 3367
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.9906,
      "step": 3368
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0002,
      "loss": 1.0423,
      "step": 3369
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 0.8512,
      "step": 3370
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.5165,
      "step": 3371
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.1294,
      "step": 3372
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.1684,
      "step": 3373
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 1.0883,
      "step": 3374
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 0.9957,
      "step": 3375
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.1928,
      "step": 3376
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.3001,
      "step": 3377
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.335,
      "step": 3378
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.4408,
      "step": 3379
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.4689,
      "step": 3380
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.0222,
      "step": 3381
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.399,
      "step": 3382
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.9913,
      "step": 3383
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.9386,
      "step": 3384
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.3519,
      "step": 3385
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 1.0578,
      "step": 3386
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 0.7631,
      "step": 3387
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0002,
      "loss": 0.4963,
      "step": 3388
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 0.4485,
      "step": 3389
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.1684,
      "step": 3390
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.3089,
      "step": 3391
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 1.0852,
      "step": 3392
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.1234,
      "step": 3393
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.3933,
      "step": 3394
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.0283,
      "step": 3395
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.7296,
      "step": 3396
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.2124,
      "step": 3397
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.0348,
      "step": 3398
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.1122,
      "step": 3399
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.1759,
      "step": 3400
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.1729,
      "step": 3401
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.8434,
      "step": 3402
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.8261,
      "step": 3403
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 0.3817,
      "step": 3404
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.1157,
      "step": 3405
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0938,
      "step": 3406
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.6363,
      "step": 3407
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.1886,
      "step": 3408
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.164,
      "step": 3409
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.3523,
      "step": 3410
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.1026,
      "step": 3411
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 0.9506,
      "step": 3412
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.895,
      "step": 3413
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.0136,
      "step": 3414
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.5332,
      "step": 3415
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.6939,
      "step": 3416
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.9415,
      "step": 3417
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.9667,
      "step": 3418
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.0026,
      "step": 3419
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.0594,
      "step": 3420
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0541,
      "step": 3421
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.3113,
      "step": 3422
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0284,
      "step": 3423
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 0.8626,
      "step": 3424
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.9646,
      "step": 3425
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 0.708,
      "step": 3426
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.2573,
      "step": 3427
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.0972,
      "step": 3428
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.1354,
      "step": 3429
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.3337,
      "step": 3430
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.2909,
      "step": 3431
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1192,
      "step": 3432
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1608,
      "step": 3433
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 0.9017,
      "step": 3434
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.4744,
      "step": 3435
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 1.1129,
      "step": 3436
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 0.9307,
      "step": 3437
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.1224,
      "step": 3438
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.258,
      "step": 3439
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.8408,
      "step": 3440
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.0949,
      "step": 3441
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.4488,
      "step": 3442
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.9823,
      "step": 3443
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0002,
      "loss": 0.4339,
      "step": 3444
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.6123,
      "step": 3445
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.6311,
      "step": 3446
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.8511,
      "step": 3447
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.1483,
      "step": 3448
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.3364,
      "step": 3449
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 1.0821,
      "step": 3450
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 0.9906,
      "step": 3451
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.2662,
      "step": 3452
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.942,
      "step": 3453
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.0524,
      "step": 3454
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.416,
      "step": 3455
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.039,
      "step": 3456
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0559,
      "step": 3457
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.9572,
      "step": 3458
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.7664,
      "step": 3459
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.3471,
      "step": 3460
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 1.4271,
      "step": 3461
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.2997,
      "step": 3462
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0711,
      "step": 3463
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 1.0333,
      "step": 3464
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.2842,
      "step": 3465
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.1605,
      "step": 3466
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.0751,
      "step": 3467
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.0521,
      "step": 3468
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.9712,
      "step": 3469
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0042,
      "step": 3470
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.0268,
      "step": 3471
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 0.9359,
      "step": 3472
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 0.8257,
      "step": 3473
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.1538,
      "step": 3474
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.1325,
      "step": 3475
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.8557,
      "step": 3476
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 0.9095,
      "step": 3477
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0002,
      "loss": 1.2423,
      "step": 3478
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 0.7036,
      "step": 3479
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.1852,
      "step": 3480
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.2165,
      "step": 3481
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.5426,
      "step": 3482
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 0.7366,
      "step": 3483
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 0.7288,
      "step": 3484
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0002,
      "loss": 0.9178,
      "step": 3485
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.3597,
      "step": 3486
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.1593,
      "step": 3487
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.1709,
      "step": 3488
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.2356,
      "step": 3489
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.8974,
      "step": 3490
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.17,
      "step": 3491
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0306,
      "step": 3492
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 0.8317,
      "step": 3493
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.3312,
      "step": 3494
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 0.9019,
      "step": 3495
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.8951,
      "step": 3496
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 1.1469,
      "step": 3497
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.3151,
      "step": 3498
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.9036,
      "step": 3499
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.0086,
      "step": 3500
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.1078,
      "step": 3501
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.2481,
      "step": 3502
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0349,
      "step": 3503
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.0468,
      "step": 3504
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.2956,
      "step": 3505
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.096,
      "step": 3506
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 0.9539,
      "step": 3507
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2097,
      "step": 3508
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.9868,
      "step": 3509
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.1772,
      "step": 3510
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.0287,
      "step": 3511
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.7047,
      "step": 3512
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.1537,
      "step": 3513
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 0.795,
      "step": 3514
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.0421,
      "step": 3515
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 1.5007,
      "step": 3516
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0002,
      "loss": 1.0876,
      "step": 3517
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.0397,
      "step": 3518
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.316,
      "step": 3519
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.2562,
      "step": 3520
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.8984,
      "step": 3521
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.0447,
      "step": 3522
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.2839,
      "step": 3523
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.7996,
      "step": 3524
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.9958,
      "step": 3525
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.3309,
      "step": 3526
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 0.6633,
      "step": 3527
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.0533,
      "step": 3528
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0002,
      "loss": 0.9509,
      "step": 3529
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.0789,
      "step": 3530
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.9944,
      "step": 3531
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.1515,
      "step": 3532
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.4604,
      "step": 3533
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.475,
      "step": 3534
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 0.8942,
      "step": 3535
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.3975,
      "step": 3536
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0002,
      "loss": 0.599,
      "step": 3537
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.6342,
      "step": 3538
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 0.8231,
      "step": 3539
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.268,
      "step": 3540
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.2436,
      "step": 3541
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.2379,
      "step": 3542
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.3004,
      "step": 3543
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 1.0107,
      "step": 3544
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 0.907,
      "step": 3545
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 0.8742,
      "step": 3546
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.1154,
      "step": 3547
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0002,
      "loss": 0.5101,
      "step": 3548
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.2879,
      "step": 3549
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.2085,
      "step": 3550
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0658,
      "step": 3551
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0956,
      "step": 3552
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.2068,
      "step": 3553
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.476,
      "step": 3554
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 0.7755,
      "step": 3555
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.1043,
      "step": 3556
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.1693,
      "step": 3557
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.8967,
      "step": 3558
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 0.7135,
      "step": 3559
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.0478,
      "step": 3560
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.5067,
      "step": 3561
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.3234,
      "step": 3562
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.2368,
      "step": 3563
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 0.755,
      "step": 3564
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.0227,
      "step": 3565
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.0461,
      "step": 3566
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.1372,
      "step": 3567
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.013,
      "step": 3568
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 0.9707,
      "step": 3569
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.4871,
      "step": 3570
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.1043,
      "step": 3571
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 0.3789,
      "step": 3572
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 0.9986,
      "step": 3573
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.0942,
      "step": 3574
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.3159,
      "step": 3575
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.3271,
      "step": 3576
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.6114,
      "step": 3577
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.0936,
      "step": 3578
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.9336,
      "step": 3579
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0918,
      "step": 3580
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 0.8425,
      "step": 3581
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 0.87,
      "step": 3582
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 0.9914,
      "step": 3583
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.0694,
      "step": 3584
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.1704,
      "step": 3585
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.2776,
      "step": 3586
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.7776,
      "step": 3587
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.8257,
      "step": 3588
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 0.9085,
      "step": 3589
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.061,
      "step": 3590
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.1349,
      "step": 3591
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.8573,
      "step": 3592
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.1777,
      "step": 3593
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1924,
      "step": 3594
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.1989,
      "step": 3595
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0002,
      "loss": 0.5849,
      "step": 3596
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.01,
      "step": 3597
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 1.0249,
      "step": 3598
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 0.9137,
      "step": 3599
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.6654,
      "step": 3600
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.1491,
      "step": 3601
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.9413,
      "step": 3602
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.4414,
      "step": 3603
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.1235,
      "step": 3604
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.0739,
      "step": 3605
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 0.9505,
      "step": 3606
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.136,
      "step": 3607
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.7825,
      "step": 3608
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.8834,
      "step": 3609
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.4509,
      "step": 3610
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.2187,
      "step": 3611
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1573,
      "step": 3612
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 0.9785,
      "step": 3613
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.2065,
      "step": 3614
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.1357,
      "step": 3615
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.9034,
      "step": 3616
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.212,
      "step": 3617
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.1323,
      "step": 3618
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.1232,
      "step": 3619
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 0.8702,
      "step": 3620
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.9383,
      "step": 3621
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.0995,
      "step": 3622
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0002,
      "loss": 1.0412,
      "step": 3623
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.1754,
      "step": 3624
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.4285,
      "step": 3625
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.367,
      "step": 3626
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.2044,
      "step": 3627
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.7184,
      "step": 3628
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.1546,
      "step": 3629
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.0577,
      "step": 3630
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 0.9539,
      "step": 3631
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.3179,
      "step": 3632
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.065,
      "step": 3633
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.0756,
      "step": 3634
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 0.5225,
      "step": 3635
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.0375,
      "step": 3636
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.3819,
      "step": 3637
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.9924,
      "step": 3638
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.9746,
      "step": 3639
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.3185,
      "step": 3640
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.2463,
      "step": 3641
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.3625,
      "step": 3642
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.195,
      "step": 3643
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.9833,
      "step": 3644
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.1205,
      "step": 3645
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.054,
      "step": 3646
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 0.4184,
      "step": 3647
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.3504,
      "step": 3648
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 0.9049,
      "step": 3649
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 0.9155,
      "step": 3650
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.8186,
      "step": 3651
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.5858,
      "step": 3652
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.3655,
      "step": 3653
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 0.584,
      "step": 3654
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.4305,
      "step": 3655
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.3714,
      "step": 3656
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.1555,
      "step": 3657
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1532,
      "step": 3658
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.2441,
      "step": 3659
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.2373,
      "step": 3660
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.022,
      "step": 3661
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.2358,
      "step": 3662
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.1414,
      "step": 3663
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.031,
      "step": 3664
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.8089,
      "step": 3665
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.0872,
      "step": 3666
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.5686,
      "step": 3667
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.1942,
      "step": 3668
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.9783,
      "step": 3669
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.7861,
      "step": 3670
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.2023,
      "step": 3671
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.3352,
      "step": 3672
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.2911,
      "step": 3673
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.0802,
      "step": 3674
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.021,
      "step": 3675
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.5986,
      "step": 3676
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.0165,
      "step": 3677
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.1511,
      "step": 3678
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.1817,
      "step": 3679
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.7738,
      "step": 3680
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.1976,
      "step": 3681
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.0523,
      "step": 3682
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 0.8763,
      "step": 3683
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.9208,
      "step": 3684
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.2791,
      "step": 3685
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.267,
      "step": 3686
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.1388,
      "step": 3687
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.4617,
      "step": 3688
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.5394,
      "step": 3689
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.0538,
      "step": 3690
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.4494,
      "step": 3691
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.4327,
      "step": 3692
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.3349,
      "step": 3693
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.2788,
      "step": 3694
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.8878,
      "step": 3695
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.1691,
      "step": 3696
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.2701,
      "step": 3697
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.175,
      "step": 3698
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.0179,
      "step": 3699
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.8656,
      "step": 3700
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 1.0213,
      "step": 3701
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.2419,
      "step": 3702
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.9737,
      "step": 3703
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.3044,
      "step": 3704
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1055,
      "step": 3705
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.2767,
      "step": 3706
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.8853,
      "step": 3707
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.7639,
      "step": 3708
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.828,
      "step": 3709
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.2525,
      "step": 3710
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.1547,
      "step": 3711
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.0836,
      "step": 3712
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 0.9249,
      "step": 3713
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 0.9622,
      "step": 3714
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.2384,
      "step": 3715
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 0.9715,
      "step": 3716
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 0.8536,
      "step": 3717
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.1187,
      "step": 3718
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.17,
      "step": 3719
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0312,
      "step": 3720
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.178,
      "step": 3721
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0406,
      "step": 3722
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.9987,
      "step": 3723
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.1293,
      "step": 3724
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.7976,
      "step": 3725
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.3904,
      "step": 3726
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 0.9037,
      "step": 3727
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.5635,
      "step": 3728
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.01,
      "step": 3729
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0002,
      "loss": 0.8301,
      "step": 3730
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.1567,
      "step": 3731
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3372,
      "step": 3732
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.8852,
      "step": 3733
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 0.7379,
      "step": 3734
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.8626,
      "step": 3735
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.146,
      "step": 3736
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.9671,
      "step": 3737
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.4531,
      "step": 3738
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.0917,
      "step": 3739
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.9452,
      "step": 3740
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.0085,
      "step": 3741
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.3343,
      "step": 3742
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.3164,
      "step": 3743
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.2474,
      "step": 3744
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.1275,
      "step": 3745
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.1232,
      "step": 3746
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.1563,
      "step": 3747
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3105,
      "step": 3748
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.2248,
      "step": 3749
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.3861,
      "step": 3750
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.0641,
      "step": 3751
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.4146,
      "step": 3752
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.8999,
      "step": 3753
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002,
      "loss": 1.4377,
      "step": 3754
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.2338,
      "step": 3755
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.81,
      "step": 3756
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.3279,
      "step": 3757
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.4217,
      "step": 3758
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 0.8378,
      "step": 3759
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.9689,
      "step": 3760
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1489,
      "step": 3761
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0148,
      "step": 3762
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 0.965,
      "step": 3763
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.4206,
      "step": 3764
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.1203,
      "step": 3765
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 1.2035,
      "step": 3766
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.1652,
      "step": 3767
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.1936,
      "step": 3768
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.5646,
      "step": 3769
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 0.9445,
      "step": 3770
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.2994,
      "step": 3771
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.81,
      "step": 3772
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.8272,
      "step": 3773
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 0.5436,
      "step": 3774
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 0.9868,
      "step": 3775
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.3047,
      "step": 3776
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.3015,
      "step": 3777
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.8408,
      "step": 3778
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 1.1689,
      "step": 3779
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3261,
      "step": 3780
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.3223,
      "step": 3781
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 0.9553,
      "step": 3782
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 0.7583,
      "step": 3783
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0889,
      "step": 3784
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.0275,
      "step": 3785
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.3842,
      "step": 3786
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.2266,
      "step": 3787
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.2166,
      "step": 3788
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.1707,
      "step": 3789
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 0.9501,
      "step": 3790
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.1042,
      "step": 3791
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.8604,
      "step": 3792
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.3437,
      "step": 3793
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.9463,
      "step": 3794
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.944,
      "step": 3795
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.8644,
      "step": 3796
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.2349,
      "step": 3797
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.8929,
      "step": 3798
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0818,
      "step": 3799
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.0926,
      "step": 3800
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.214,
      "step": 3801
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.1003,
      "step": 3802
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.0594,
      "step": 3803
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.3587,
      "step": 3804
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.1323,
      "step": 3805
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 1.0549,
      "step": 3806
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.1161,
      "step": 3807
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.3479,
      "step": 3808
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.4767,
      "step": 3809
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.3492,
      "step": 3810
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 0.8519,
      "step": 3811
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 0.8996,
      "step": 3812
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.0922,
      "step": 3813
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.701,
      "step": 3814
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.2198,
      "step": 3815
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.2131,
      "step": 3816
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.3256,
      "step": 3817
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.0306,
      "step": 3818
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.1552,
      "step": 3819
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 0.8993,
      "step": 3820
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2266,
      "step": 3821
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 1.0263,
      "step": 3822
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.2948,
      "step": 3823
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 0.9176,
      "step": 3824
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 0.6847,
      "step": 3825
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.402,
      "step": 3826
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.778,
      "step": 3827
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.3656,
      "step": 3828
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.5472,
      "step": 3829
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.9963,
      "step": 3830
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0373,
      "step": 3831
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.1217,
      "step": 3832
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.161,
      "step": 3833
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2463,
      "step": 3834
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.1016,
      "step": 3835
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.4005,
      "step": 3836
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.9707,
      "step": 3837
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.4573,
      "step": 3838
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 0.9813,
      "step": 3839
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 0.9968,
      "step": 3840
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.0225,
      "step": 3841
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.0941,
      "step": 3842
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.0735,
      "step": 3843
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.9293,
      "step": 3844
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.5347,
      "step": 3845
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.1304,
      "step": 3846
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.0529,
      "step": 3847
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.4404,
      "step": 3848
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 0.9736,
      "step": 3849
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.2002,
      "step": 3850
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 0.836,
      "step": 3851
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.0733,
      "step": 3852
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 0.9964,
      "step": 3853
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 1.2169,
      "step": 3854
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.2655,
      "step": 3855
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 0.891,
      "step": 3856
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0081,
      "step": 3857
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.2464,
      "step": 3858
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.9119,
      "step": 3859
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.7645,
      "step": 3860
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.1547,
      "step": 3861
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.6264,
      "step": 3862
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2535,
      "step": 3863
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.069,
      "step": 3864
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.1939,
      "step": 3865
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.0828,
      "step": 3866
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 0.8978,
      "step": 3867
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.1434,
      "step": 3868
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.9916,
      "step": 3869
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.2974,
      "step": 3870
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.7487,
      "step": 3871
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.3163,
      "step": 3872
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.1742,
      "step": 3873
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.1328,
      "step": 3874
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.0415,
      "step": 3875
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.4527,
      "step": 3876
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 0.4978,
      "step": 3877
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.1824,
      "step": 3878
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.2578,
      "step": 3879
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0509,
      "step": 3880
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.1983,
      "step": 3881
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.1224,
      "step": 3882
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.1806,
      "step": 3883
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1256,
      "step": 3884
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.9255,
      "step": 3885
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.3043,
      "step": 3886
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.2476,
      "step": 3887
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.3129,
      "step": 3888
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.262,
      "step": 3889
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.4661,
      "step": 3890
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.4309,
      "step": 3891
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.0519,
      "step": 3892
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0002,
      "loss": 0.8654,
      "step": 3893
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.8453,
      "step": 3894
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.0917,
      "step": 3895
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.3197,
      "step": 3896
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.0616,
      "step": 3897
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1861,
      "step": 3898
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.4221,
      "step": 3899
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.4361,
      "step": 3900
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.9567,
      "step": 3901
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.106,
      "step": 3902
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.6333,
      "step": 3903
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2403,
      "step": 3904
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.3075,
      "step": 3905
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 0.9298,
      "step": 3906
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.3373,
      "step": 3907
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.0896,
      "step": 3908
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.0728,
      "step": 3909
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.1635,
      "step": 3910
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.3475,
      "step": 3911
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 1.2174,
      "step": 3912
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.0574,
      "step": 3913
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0002,
      "loss": 1.1515,
      "step": 3914
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.4558,
      "step": 3915
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.0667,
      "step": 3916
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 0.951,
      "step": 3917
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.1914,
      "step": 3918
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 0.9842,
      "step": 3919
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0002,
      "loss": 0.9518,
      "step": 3920
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.9903,
      "step": 3921
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.4631,
      "step": 3922
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.0758,
      "step": 3923
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 0.8558,
      "step": 3924
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.0017,
      "step": 3925
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0002,
      "loss": 0.8005,
      "step": 3926
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.4049,
      "step": 3927
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.3374,
      "step": 3928
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.2417,
      "step": 3929
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.1707,
      "step": 3930
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.206,
      "step": 3931
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.4059,
      "step": 3932
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.0595,
      "step": 3933
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.9897,
      "step": 3934
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.0938,
      "step": 3935
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.1353,
      "step": 3936
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.4123,
      "step": 3937
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.9125,
      "step": 3938
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.0832,
      "step": 3939
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1144,
      "step": 3940
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.8653,
      "step": 3941
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.025,
      "step": 3942
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 0.7614,
      "step": 3943
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.1325,
      "step": 3944
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.9262,
      "step": 3945
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.1181,
      "step": 3946
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 0.8408,
      "step": 3947
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.9908,
      "step": 3948
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.0314,
      "step": 3949
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.2201,
      "step": 3950
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.2811,
      "step": 3951
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.2286,
      "step": 3952
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.065,
      "step": 3953
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.3893,
      "step": 3954
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.2077,
      "step": 3955
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.3064,
      "step": 3956
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.0489,
      "step": 3957
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.1013,
      "step": 3958
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.7481,
      "step": 3959
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.2781,
      "step": 3960
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 0.9432,
      "step": 3961
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.9077,
      "step": 3962
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.149,
      "step": 3963
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.9011,
      "step": 3964
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.3677,
      "step": 3965
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.3748,
      "step": 3966
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.1496,
      "step": 3967
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.6264,
      "step": 3968
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0002,
      "loss": 1.143,
      "step": 3969
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.2643,
      "step": 3970
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.1792,
      "step": 3971
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.0277,
      "step": 3972
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 0.8804,
      "step": 3973
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.9144,
      "step": 3974
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.9586,
      "step": 3975
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.0892,
      "step": 3976
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.0487,
      "step": 3977
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.0794,
      "step": 3978
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.3631,
      "step": 3979
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.2802,
      "step": 3980
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.276,
      "step": 3981
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 0.993,
      "step": 3982
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.1926,
      "step": 3983
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.1547,
      "step": 3984
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 0.9023,
      "step": 3985
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.7777,
      "step": 3986
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.086,
      "step": 3987
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.2924,
      "step": 3988
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.2705,
      "step": 3989
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.0794,
      "step": 3990
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 0.9607,
      "step": 3991
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.3763,
      "step": 3992
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.3797,
      "step": 3993
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.9532,
      "step": 3994
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.4232,
      "step": 3995
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.0905,
      "step": 3996
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.7755,
      "step": 3997
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.2814,
      "step": 3998
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.8705,
      "step": 3999
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 0.5414,
      "step": 4000
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 1.1129,
      "step": 4001
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.4263,
      "step": 4002
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.5855,
      "step": 4003
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 1.0643,
      "step": 4004
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.1679,
      "step": 4005
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.2028,
      "step": 4006
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 0.7004,
      "step": 4007
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0002,
      "loss": 1.0713,
      "step": 4008
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.2463,
      "step": 4009
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.1266,
      "step": 4010
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.086,
      "step": 4011
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.7535,
      "step": 4012
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.142,
      "step": 4013
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.5189,
      "step": 4014
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 1.161,
      "step": 4015
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.1645,
      "step": 4016
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.9177,
      "step": 4017
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.9686,
      "step": 4018
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.4932,
      "step": 4019
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.885,
      "step": 4020
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 0.7899,
      "step": 4021
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.5013,
      "step": 4022
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.1371,
      "step": 4023
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.1725,
      "step": 4024
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 1.1008,
      "step": 4025
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.9327,
      "step": 4026
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.2686,
      "step": 4027
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.9923,
      "step": 4028
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.0615,
      "step": 4029
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.1571,
      "step": 4030
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.0116,
      "step": 4031
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.9315,
      "step": 4032
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.2037,
      "step": 4033
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.9035,
      "step": 4034
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.9755,
      "step": 4035
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.1186,
      "step": 4036
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 0.9367,
      "step": 4037
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.4776,
      "step": 4038
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.031,
      "step": 4039
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.2725,
      "step": 4040
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.7716,
      "step": 4041
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.2188,
      "step": 4042
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.486,
      "step": 4043
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.1378,
      "step": 4044
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 0.9148,
      "step": 4045
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.3427,
      "step": 4046
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.3768,
      "step": 4047
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.1626,
      "step": 4048
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0629,
      "step": 4049
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.3198,
      "step": 4050
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.0569,
      "step": 4051
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1254,
      "step": 4052
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.2101,
      "step": 4053
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 1.1992,
      "step": 4054
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.0138,
      "step": 4055
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 0.5812,
      "step": 4056
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.9476,
      "step": 4057
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.2213,
      "step": 4058
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.9002,
      "step": 4059
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.1541,
      "step": 4060
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 0.9007,
      "step": 4061
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3229,
      "step": 4062
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.3606,
      "step": 4063
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.1814,
      "step": 4064
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.0316,
      "step": 4065
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 0.8485,
      "step": 4066
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.6441,
      "step": 4067
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.0894,
      "step": 4068
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.6763,
      "step": 4069
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 0.7946,
      "step": 4070
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 0.8548,
      "step": 4071
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.8705,
      "step": 4072
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.8826,
      "step": 4073
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.2292,
      "step": 4074
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 0.8053,
      "step": 4075
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1743,
      "step": 4076
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 0.9134,
      "step": 4077
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.2348,
      "step": 4078
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.1057,
      "step": 4079
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.4598,
      "step": 4080
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.1696,
      "step": 4081
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 0.3678,
      "step": 4082
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.9436,
      "step": 4083
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.2413,
      "step": 4084
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.1981,
      "step": 4085
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.2223,
      "step": 4086
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 0.8099,
      "step": 4087
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0947,
      "step": 4088
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.0644,
      "step": 4089
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.7296,
      "step": 4090
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 1.1077,
      "step": 4091
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.389,
      "step": 4092
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 0.8372,
      "step": 4093
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.0814,
      "step": 4094
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.2921,
      "step": 4095
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.871,
      "step": 4096
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.2157,
      "step": 4097
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.0092,
      "step": 4098
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.1898,
      "step": 4099
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.2636,
      "step": 4100
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.2794,
      "step": 4101
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.3134,
      "step": 4102
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.2267,
      "step": 4103
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 0.9569,
      "step": 4104
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.1289,
      "step": 4105
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.2847,
      "step": 4106
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.4477,
      "step": 4107
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 0.9664,
      "step": 4108
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.0621,
      "step": 4109
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.0003,
      "step": 4110
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.4913,
      "step": 4111
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.4274,
      "step": 4112
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.0056,
      "step": 4113
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.2564,
      "step": 4114
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.3233,
      "step": 4115
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.1467,
      "step": 4116
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0002,
      "loss": 0.9309,
      "step": 4117
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.2237,
      "step": 4118
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.2583,
      "step": 4119
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 0.9681,
      "step": 4120
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 0.9748,
      "step": 4121
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2382,
      "step": 4122
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.9566,
      "step": 4123
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 0.9254,
      "step": 4124
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2647,
      "step": 4125
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 0.8175,
      "step": 4126
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 0.9945,
      "step": 4127
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.0649,
      "step": 4128
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.0167,
      "step": 4129
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0907,
      "step": 4130
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.2387,
      "step": 4131
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.0789,
      "step": 4132
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1319,
      "step": 4133
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.5844,
      "step": 4134
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.9632,
      "step": 4135
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.1826,
      "step": 4136
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.0049,
      "step": 4137
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.8743,
      "step": 4138
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.0567,
      "step": 4139
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.905,
      "step": 4140
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.0331,
      "step": 4141
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 0.9786,
      "step": 4142
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.3898,
      "step": 4143
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.5376,
      "step": 4144
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.2452,
      "step": 4145
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 0.7851,
      "step": 4146
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2814,
      "step": 4147
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 0.8724,
      "step": 4148
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.3989,
      "step": 4149
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.2261,
      "step": 4150
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 0.9412,
      "step": 4151
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.5528,
      "step": 4152
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.3038,
      "step": 4153
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.462,
      "step": 4154
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0483,
      "step": 4155
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.3097,
      "step": 4156
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.3068,
      "step": 4157
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.9292,
      "step": 4158
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1163,
      "step": 4159
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.5268,
      "step": 4160
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.8286,
      "step": 4161
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.0288,
      "step": 4162
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.255,
      "step": 4163
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 0.9084,
      "step": 4164
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.2857,
      "step": 4165
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.6546,
      "step": 4166
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 0.9473,
      "step": 4167
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.5638,
      "step": 4168
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.2477,
      "step": 4169
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.2323,
      "step": 4170
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.9992,
      "step": 4171
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 0.8281,
      "step": 4172
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.124,
      "step": 4173
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.368,
      "step": 4174
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.2319,
      "step": 4175
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 0.6795,
      "step": 4176
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2239,
      "step": 4177
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 0.7322,
      "step": 4178
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.0553,
      "step": 4179
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.1719,
      "step": 4180
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.3873,
      "step": 4181
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.2605,
      "step": 4182
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.1378,
      "step": 4183
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.4291,
      "step": 4184
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.0117,
      "step": 4185
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.191,
      "step": 4186
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.0898,
      "step": 4187
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.4175,
      "step": 4188
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.1321,
      "step": 4189
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.9641,
      "step": 4190
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.0719,
      "step": 4191
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 0.9778,
      "step": 4192
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.3273,
      "step": 4193
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.6223,
      "step": 4194
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.0704,
      "step": 4195
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.7496,
      "step": 4196
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 1.2488,
      "step": 4197
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 0.9209,
      "step": 4198
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.1404,
      "step": 4199
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.2642,
      "step": 4200
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.0251,
      "step": 4201
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.0937,
      "step": 4202
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.3024,
      "step": 4203
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.5073,
      "step": 4204
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 0.777,
      "step": 4205
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.2155,
      "step": 4206
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.0556,
      "step": 4207
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.196,
      "step": 4208
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.2047,
      "step": 4209
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.2365,
      "step": 4210
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.1116,
      "step": 4211
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0002,
      "loss": 1.0502,
      "step": 4212
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.1508,
      "step": 4213
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.3639,
      "step": 4214
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.2839,
      "step": 4215
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.3832,
      "step": 4216
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.3346,
      "step": 4217
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.1485,
      "step": 4218
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.2918,
      "step": 4219
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.5917,
      "step": 4220
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.9753,
      "step": 4221
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.2868,
      "step": 4222
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.437,
      "step": 4223
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0921,
      "step": 4224
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.2872,
      "step": 4225
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.3018,
      "step": 4226
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.397,
      "step": 4227
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.4951,
      "step": 4228
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.3222,
      "step": 4229
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 1.2421,
      "step": 4230
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.0307,
      "step": 4231
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.3908,
      "step": 4232
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 0.9011,
      "step": 4233
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.1614,
      "step": 4234
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.0733,
      "step": 4235
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 0.9876,
      "step": 4236
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0239,
      "step": 4237
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 0.8424,
      "step": 4238
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.0152,
      "step": 4239
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.2604,
      "step": 4240
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.1758,
      "step": 4241
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.2924,
      "step": 4242
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.4813,
      "step": 4243
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 0.987,
      "step": 4244
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.2415,
      "step": 4245
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.0565,
      "step": 4246
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.0116,
      "step": 4247
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.5158,
      "step": 4248
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2198,
      "step": 4249
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 0.9393,
      "step": 4250
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.1915,
      "step": 4251
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.9251,
      "step": 4252
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.2436,
      "step": 4253
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.3713,
      "step": 4254
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.2075,
      "step": 4255
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 1.0131,
      "step": 4256
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.1003,
      "step": 4257
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.082,
      "step": 4258
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2058,
      "step": 4259
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.0661,
      "step": 4260
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.8309,
      "step": 4261
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3802,
      "step": 4262
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.4087,
      "step": 4263
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.5365,
      "step": 4264
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.5394,
      "step": 4265
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1005,
      "step": 4266
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.1559,
      "step": 4267
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.24,
      "step": 4268
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 0.9033,
      "step": 4269
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.4306,
      "step": 4270
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.2497,
      "step": 4271
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 0.846,
      "step": 4272
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 1.1618,
      "step": 4273
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.1446,
      "step": 4274
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.5038,
      "step": 4275
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.0041,
      "step": 4276
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 1.1251,
      "step": 4277
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.2418,
      "step": 4278
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.1662,
      "step": 4279
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.9206,
      "step": 4280
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.4513,
      "step": 4281
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.45,
      "step": 4282
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.3373,
      "step": 4283
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.5365,
      "step": 4284
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 0.9499,
      "step": 4285
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.2437,
      "step": 4286
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.7669,
      "step": 4287
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.9113,
      "step": 4288
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.4465,
      "step": 4289
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.784,
      "step": 4290
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.0927,
      "step": 4291
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.4082,
      "step": 4292
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.2357,
      "step": 4293
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 1.0217,
      "step": 4294
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.6015,
      "step": 4295
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.0282,
      "step": 4296
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 0.9909,
      "step": 4297
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.0774,
      "step": 4298
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.3297,
      "step": 4299
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.0111,
      "step": 4300
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.0668,
      "step": 4301
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.6019,
      "step": 4302
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.2988,
      "step": 4303
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.3808,
      "step": 4304
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0002,
      "loss": 1.4143,
      "step": 4305
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.133,
      "step": 4306
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.2793,
      "step": 4307
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.554,
      "step": 4308
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.1504,
      "step": 4309
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.9605,
      "step": 4310
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.981,
      "step": 4311
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.3866,
      "step": 4312
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.168,
      "step": 4313
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.0254,
      "step": 4314
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3975,
      "step": 4315
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.3589,
      "step": 4316
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.3599,
      "step": 4317
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.258,
      "step": 4318
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 0.7849,
      "step": 4319
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.2335,
      "step": 4320
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.9961,
      "step": 4321
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.6925,
      "step": 4322
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.4294,
      "step": 4323
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.8811,
      "step": 4324
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.4473,
      "step": 4325
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0475,
      "step": 4326
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 0.9215,
      "step": 4327
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.2999,
      "step": 4328
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.1877,
      "step": 4329
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.0546,
      "step": 4330
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 0.9765,
      "step": 4331
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.0822,
      "step": 4332
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.938,
      "step": 4333
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.2581,
      "step": 4334
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.0544,
      "step": 4335
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0002,
      "loss": 0.9464,
      "step": 4336
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.1265,
      "step": 4337
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.0329,
      "step": 4338
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.1616,
      "step": 4339
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.7737,
      "step": 4340
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 0.9756,
      "step": 4341
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1353,
      "step": 4342
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0522,
      "step": 4343
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0587,
      "step": 4344
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.6962,
      "step": 4345
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.0195,
      "step": 4346
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 1.3268,
      "step": 4347
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.0512,
      "step": 4348
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 0.8434,
      "step": 4349
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.1639,
      "step": 4350
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.5723,
      "step": 4351
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.3958,
      "step": 4352
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.9973,
      "step": 4353
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.9294,
      "step": 4354
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.9895,
      "step": 4355
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.9493,
      "step": 4356
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.0327,
      "step": 4357
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.1486,
      "step": 4358
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.09716796875,
      "learning_rate": 0.0002,
      "loss": 0.3253,
      "step": 4359
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.1343,
      "step": 4360
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.518,
      "step": 4361
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.5683,
      "step": 4362
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.1388,
      "step": 4363
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 0.7407,
      "step": 4364
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 1.2496,
      "step": 4365
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.1549,
      "step": 4366
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.0273,
      "step": 4367
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.2719,
      "step": 4368
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1111,
      "step": 4369
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0963,
      "step": 4370
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.028,
      "step": 4371
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.2837,
      "step": 4372
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.2466,
      "step": 4373
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 0.9697,
      "step": 4374
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 1.4277,
      "step": 4375
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.2217,
      "step": 4376
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3276,
      "step": 4377
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.4667,
      "step": 4378
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.0638,
      "step": 4379
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.0182,
      "step": 4380
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.3033,
      "step": 4381
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0002,
      "loss": 0.4494,
      "step": 4382
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.1367,
      "step": 4383
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.565,
      "step": 4384
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.158,
      "step": 4385
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.0121,
      "step": 4386
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.8617,
      "step": 4387
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.2633,
      "step": 4388
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.5808,
      "step": 4389
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 0.8747,
      "step": 4390
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.317,
      "step": 4391
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.4163,
      "step": 4392
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.0518,
      "step": 4393
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.3643,
      "step": 4394
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.4018,
      "step": 4395
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 0.9976,
      "step": 4396
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.9978,
      "step": 4397
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.1158,
      "step": 4398
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.0291,
      "step": 4399
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.1415,
      "step": 4400
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0065,
      "step": 4401
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.0076,
      "step": 4402
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.9042,
      "step": 4403
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 0.8928,
      "step": 4404
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.901,
      "step": 4405
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.5914,
      "step": 4406
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.4093,
      "step": 4407
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.7541,
      "step": 4408
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.6444,
      "step": 4409
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.0752,
      "step": 4410
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.553,
      "step": 4411
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.0352,
      "step": 4412
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.984,
      "step": 4413
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 1.0518,
      "step": 4414
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.0919,
      "step": 4415
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 0.9291,
      "step": 4416
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 0.6598,
      "step": 4417
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.0077,
      "step": 4418
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.415,
      "step": 4419
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.1628,
      "step": 4420
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.1672,
      "step": 4421
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.2645,
      "step": 4422
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0822,
      "step": 4423
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.2107,
      "step": 4424
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.5491,
      "step": 4425
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 1.0458,
      "step": 4426
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1747,
      "step": 4427
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.9439,
      "step": 4428
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.9525,
      "step": 4429
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.5354,
      "step": 4430
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.9948,
      "step": 4431
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0002,
      "loss": 0.7726,
      "step": 4432
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.101,
      "step": 4433
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.0238,
      "step": 4434
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.3717,
      "step": 4435
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 0.6927,
      "step": 4436
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2517,
      "step": 4437
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.8936,
      "step": 4438
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.8328,
      "step": 4439
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.6954,
      "step": 4440
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.0049,
      "step": 4441
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.3157,
      "step": 4442
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.3329,
      "step": 4443
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.3391,
      "step": 4444
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.0426,
      "step": 4445
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.1001,
      "step": 4446
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.9655,
      "step": 4447
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 0.9577,
      "step": 4448
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 0.4827,
      "step": 4449
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.2696,
      "step": 4450
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.2518,
      "step": 4451
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.0417,
      "step": 4452
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.2908,
      "step": 4453
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.0157,
      "step": 4454
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.1858,
      "step": 4455
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.9022,
      "step": 4456
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2385,
      "step": 4457
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.3089,
      "step": 4458
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.9152,
      "step": 4459
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.9648,
      "step": 4460
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.2107,
      "step": 4461
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.0399,
      "step": 4462
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.8905,
      "step": 4463
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.0157,
      "step": 4464
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3444,
      "step": 4465
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.3286,
      "step": 4466
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.148,
      "step": 4467
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.0867,
      "step": 4468
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.9349,
      "step": 4469
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.2464,
      "step": 4470
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.2033,
      "step": 4471
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.958,
      "step": 4472
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 1.1403,
      "step": 4473
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.3084,
      "step": 4474
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 0.8739,
      "step": 4475
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3408,
      "step": 4476
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.5679,
      "step": 4477
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.0014,
      "step": 4478
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.0042,
      "step": 4479
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 0.9329,
      "step": 4480
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.206,
      "step": 4481
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 0.9705,
      "step": 4482
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.0001,
      "step": 4483
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.83,
      "step": 4484
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.1197,
      "step": 4485
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.8527,
      "step": 4486
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.256,
      "step": 4487
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.2399,
      "step": 4488
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.0823,
      "step": 4489
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.0142,
      "step": 4490
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.3808,
      "step": 4491
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.3456,
      "step": 4492
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1005859375,
      "learning_rate": 0.0002,
      "loss": 0.5637,
      "step": 4493
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.247,
      "step": 4494
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 0.9808,
      "step": 4495
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.1837,
      "step": 4496
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 0.6745,
      "step": 4497
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 0.9894,
      "step": 4498
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1399,
      "step": 4499
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.1488,
      "step": 4500
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.2905,
      "step": 4501
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 0.5526,
      "step": 4502
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.0137,
      "step": 4503
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.9303,
      "step": 4504
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.0856,
      "step": 4505
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.4775,
      "step": 4506
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.1213,
      "step": 4507
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.1422,
      "step": 4508
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.5922,
      "step": 4509
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 0.9907,
      "step": 4510
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.0258,
      "step": 4511
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.5149,
      "step": 4512
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.4093,
      "step": 4513
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.0545,
      "step": 4514
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.1461,
      "step": 4515
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0002,
      "loss": 0.8313,
      "step": 4516
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.1207,
      "step": 4517
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.0945,
      "step": 4518
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.3624,
      "step": 4519
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.4262,
      "step": 4520
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.1141,
      "step": 4521
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.1321,
      "step": 4522
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 1.1454,
      "step": 4523
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.0304,
      "step": 4524
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.0734,
      "step": 4525
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 0.828,
      "step": 4526
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.3388,
      "step": 4527
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1934,
      "step": 4528
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.7939,
      "step": 4529
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.0223,
      "step": 4530
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.0859,
      "step": 4531
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.2405,
      "step": 4532
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.1397,
      "step": 4533
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.2544,
      "step": 4534
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1844,
      "step": 4535
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.1212,
      "step": 4536
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.2407,
      "step": 4537
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.0839,
      "step": 4538
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.076,
      "step": 4539
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.7097,
      "step": 4540
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.3707,
      "step": 4541
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.294,
      "step": 4542
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 0.3952,
      "step": 4543
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3134,
      "step": 4544
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.9211,
      "step": 4545
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 0.8746,
      "step": 4546
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.6988,
      "step": 4547
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.4102,
      "step": 4548
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.0277,
      "step": 4549
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 1.4081,
      "step": 4550
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.186,
      "step": 4551
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.1703,
      "step": 4552
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.1541,
      "step": 4553
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.8996,
      "step": 4554
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.374,
      "step": 4555
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 0.9933,
      "step": 4556
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.2271,
      "step": 4557
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.2506,
      "step": 4558
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.3312,
      "step": 4559
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.4321,
      "step": 4560
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0484,
      "step": 4561
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0211,
      "step": 4562
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.964,
      "step": 4563
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.2468,
      "step": 4564
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.2375,
      "step": 4565
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.4175,
      "step": 4566
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.3808,
      "step": 4567
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.7732,
      "step": 4568
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.0306,
      "step": 4569
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.031,
      "step": 4570
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 1.092,
      "step": 4571
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.9705,
      "step": 4572
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.0082,
      "step": 4573
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.1832,
      "step": 4574
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0002,
      "loss": 1.1415,
      "step": 4575
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.1998,
      "step": 4576
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0399,
      "step": 4577
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.9506,
      "step": 4578
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.0182,
      "step": 4579
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.4535,
      "step": 4580
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 0.801,
      "step": 4581
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.3473,
      "step": 4582
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 0.8195,
      "step": 4583
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 0.6467,
      "step": 4584
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.3214,
      "step": 4585
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 1.3381,
      "step": 4586
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.1636,
      "step": 4587
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.9222,
      "step": 4588
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.0383,
      "step": 4589
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 1.0327,
      "step": 4590
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.1759,
      "step": 4591
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.1196,
      "step": 4592
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 0.9517,
      "step": 4593
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 0.9618,
      "step": 4594
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.248,
      "step": 4595
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.9567,
      "step": 4596
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 0.7683,
      "step": 4597
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.1403,
      "step": 4598
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.3583,
      "step": 4599
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0602,
      "step": 4600
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.0423,
      "step": 4601
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.3647,
      "step": 4602
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.9299,
      "step": 4603
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0002,
      "loss": 1.0799,
      "step": 4604
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 0.6868,
      "step": 4605
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 0.8907,
      "step": 4606
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.0341,
      "step": 4607
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.1301,
      "step": 4608
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.0097,
      "step": 4609
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.487,
      "step": 4610
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1475,
      "step": 4611
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.1485,
      "step": 4612
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.5035,
      "step": 4613
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.186,
      "step": 4614
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.0232,
      "step": 4615
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.3527,
      "step": 4616
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.4297,
      "step": 4617
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 0.5979,
      "step": 4618
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0274,
      "step": 4619
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.2689,
      "step": 4620
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 1.0846,
      "step": 4621
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.0461,
      "step": 4622
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.991,
      "step": 4623
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.2555,
      "step": 4624
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.2423,
      "step": 4625
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.4466,
      "step": 4626
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.1858,
      "step": 4627
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.5552,
      "step": 4628
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.1225,
      "step": 4629
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.51,
      "step": 4630
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.1036,
      "step": 4631
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.0155,
      "step": 4632
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.9939,
      "step": 4633
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.0364,
      "step": 4634
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 0.7855,
      "step": 4635
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 0.8453,
      "step": 4636
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.0562,
      "step": 4637
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.9794,
      "step": 4638
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.3345,
      "step": 4639
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2449,
      "step": 4640
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 1.2728,
      "step": 4641
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.8028,
      "step": 4642
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.063,
      "step": 4643
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.8673,
      "step": 4644
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.2915,
      "step": 4645
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.9708,
      "step": 4646
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0002,
      "loss": 0.6871,
      "step": 4647
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.8981,
      "step": 4648
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2384,
      "step": 4649
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.5911,
      "step": 4650
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0002,
      "loss": 0.777,
      "step": 4651
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.901,
      "step": 4652
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 0.8055,
      "step": 4653
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.3811,
      "step": 4654
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.6359,
      "step": 4655
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.6484,
      "step": 4656
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.2401,
      "step": 4657
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 0.9405,
      "step": 4658
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 1.3727,
      "step": 4659
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.3834,
      "step": 4660
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.4413,
      "step": 4661
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.803,
      "step": 4662
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.2012,
      "step": 4663
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.231,
      "step": 4664
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.211,
      "step": 4665
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.1263,
      "step": 4666
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.3179,
      "step": 4667
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.3492,
      "step": 4668
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.3205,
      "step": 4669
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 0.9563,
      "step": 4670
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.233,
      "step": 4671
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.3893,
      "step": 4672
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.1421,
      "step": 4673
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.327,
      "step": 4674
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.0715,
      "step": 4675
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.0329,
      "step": 4676
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 0.9221,
      "step": 4677
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.0127,
      "step": 4678
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.2081,
      "step": 4679
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.1067,
      "step": 4680
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.2011,
      "step": 4681
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.105,
      "step": 4682
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.9913,
      "step": 4683
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 0.9552,
      "step": 4684
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.0803,
      "step": 4685
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2242,
      "step": 4686
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.3376,
      "step": 4687
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.6474,
      "step": 4688
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.5407,
      "step": 4689
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 0.9712,
      "step": 4690
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.3176,
      "step": 4691
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.2266,
      "step": 4692
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.0206,
      "step": 4693
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.0576,
      "step": 4694
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.2021,
      "step": 4695
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.3381,
      "step": 4696
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 0.9604,
      "step": 4697
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.8728,
      "step": 4698
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.07,
      "step": 4699
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.2874,
      "step": 4700
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.2303,
      "step": 4701
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.7695,
      "step": 4702
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.2277,
      "step": 4703
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.1087,
      "step": 4704
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.8771,
      "step": 4705
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.2656,
      "step": 4706
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.5965,
      "step": 4707
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.2165,
      "step": 4708
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 0.8933,
      "step": 4709
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.4353,
      "step": 4710
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.0438,
      "step": 4711
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.1106,
      "step": 4712
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.2681,
      "step": 4713
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 0.9073,
      "step": 4714
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.1625,
      "step": 4715
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0002,
      "loss": 0.8804,
      "step": 4716
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.9201,
      "step": 4717
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0002,
      "loss": 0.3052,
      "step": 4718
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.1187,
      "step": 4719
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1178,
      "step": 4720
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.9673,
      "step": 4721
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.3526,
      "step": 4722
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 0.9667,
      "step": 4723
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.1158,
      "step": 4724
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.8027,
      "step": 4725
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.8223,
      "step": 4726
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.8656,
      "step": 4727
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 1.2862,
      "step": 4728
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.5004,
      "step": 4729
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.0899,
      "step": 4730
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.8087,
      "step": 4731
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0002,
      "loss": 1.182,
      "step": 4732
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.6571,
      "step": 4733
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.1766,
      "step": 4734
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.8317,
      "step": 4735
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.0394,
      "step": 4736
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.7936,
      "step": 4737
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.09130859375,
      "learning_rate": 0.0002,
      "loss": 0.3855,
      "step": 4738
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.9082,
      "step": 4739
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.8753,
      "step": 4740
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 0.8938,
      "step": 4741
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.3811,
      "step": 4742
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0002,
      "loss": 0.8414,
      "step": 4743
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0714,
      "step": 4744
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1201,
      "step": 4745
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.2214,
      "step": 4746
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.3543,
      "step": 4747
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.9593,
      "step": 4748
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.2862,
      "step": 4749
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 0.9641,
      "step": 4750
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 0.8684,
      "step": 4751
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.7857,
      "step": 4752
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.0115,
      "step": 4753
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.9089,
      "step": 4754
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 0.952,
      "step": 4755
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.486,
      "step": 4756
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.0867,
      "step": 4757
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 0.8198,
      "step": 4758
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0002,
      "loss": 0.914,
      "step": 4759
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.2514,
      "step": 4760
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.0981,
      "step": 4761
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.3299,
      "step": 4762
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.1901,
      "step": 4763
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 0.8484,
      "step": 4764
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.2687,
      "step": 4765
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.4107,
      "step": 4766
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.5561,
      "step": 4767
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 0.5041,
      "step": 4768
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.474,
      "step": 4769
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.1231,
      "step": 4770
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.4913,
      "step": 4771
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.0268,
      "step": 4772
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 0.7797,
      "step": 4773
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.1227,
      "step": 4774
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.3301,
      "step": 4775
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.1853,
      "step": 4776
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.9143,
      "step": 4777
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.9394,
      "step": 4778
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0147,
      "step": 4779
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 1.3655,
      "step": 4780
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.3528,
      "step": 4781
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 0.8042,
      "step": 4782
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.2909,
      "step": 4783
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.2978,
      "step": 4784
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.0515,
      "step": 4785
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.5625,
      "learning_rate": 0.0002,
      "loss": 1.4234,
      "step": 4786
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 0.8757,
      "step": 4787
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.2333,
      "step": 4788
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.1655,
      "step": 4789
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.1253,
      "step": 4790
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.0744,
      "step": 4791
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 0.9071,
      "step": 4792
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.0709,
      "step": 4793
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 0.9893,
      "step": 4794
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.2815,
      "step": 4795
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.6361,
      "step": 4796
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.2823,
      "step": 4797
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.2488,
      "step": 4798
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0002,
      "loss": 1.058,
      "step": 4799
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0366,
      "step": 4800
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.389,
      "step": 4801
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.0111,
      "step": 4802
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.9797,
      "step": 4803
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.2336,
      "step": 4804
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.1809,
      "step": 4805
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.2733,
      "step": 4806
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.1454,
      "step": 4807
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.8235,
      "step": 4808
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.489,
      "step": 4809
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.1981,
      "step": 4810
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.1239,
      "step": 4811
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.4597,
      "step": 4812
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.2099,
      "step": 4813
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.2508,
      "step": 4814
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.353,
      "step": 4815
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 0.9032,
      "step": 4816
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 0.5823,
      "step": 4817
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.734375,
      "learning_rate": 0.0002,
      "loss": 0.8317,
      "step": 4818
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.0444,
      "step": 4819
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 0.9934,
      "step": 4820
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 1.1316,
      "step": 4821
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.1823,
      "step": 4822
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.2831,
      "step": 4823
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 0.9461,
      "step": 4824
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.9321,
      "step": 4825
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.2535,
      "step": 4826
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.0904,
      "step": 4827
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 0.9893,
      "step": 4828
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.0154,
      "step": 4829
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.2263,
      "step": 4830
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.073,
      "step": 4831
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.0835,
      "step": 4832
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1909,
      "step": 4833
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.3459,
      "step": 4834
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 0.7457,
      "step": 4835
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 0.982,
      "step": 4836
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 0.8051,
      "step": 4837
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.0006,
      "step": 4838
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.2495,
      "step": 4839
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.3136,
      "step": 4840
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.3941,
      "step": 4841
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.0348,
      "step": 4842
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.2482,
      "step": 4843
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.9833,
      "step": 4844
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.2315,
      "step": 4845
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.3635,
      "step": 4846
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 0.9956,
      "step": 4847
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.232,
      "step": 4848
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.4683,
      "step": 4849
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.5042,
      "step": 4850
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.0333,
      "step": 4851
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1765,
      "step": 4852
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.4089,
      "step": 4853
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.0562,
      "step": 4854
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.2057,
      "step": 4855
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.1658,
      "step": 4856
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.1323,
      "step": 4857
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.4792,
      "step": 4858
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.6398,
      "step": 4859
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.418,
      "step": 4860
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.2033,
      "step": 4861
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.129,
      "step": 4862
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 0.6353,
      "step": 4863
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.0405,
      "step": 4864
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.3031,
      "step": 4865
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.0433,
      "step": 4866
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.0333,
      "step": 4867
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.2574,
      "step": 4868
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.0298,
      "step": 4869
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.7676,
      "step": 4870
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.3453,
      "step": 4871
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 0.8259,
      "step": 4872
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.9812,
      "step": 4873
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.226,
      "step": 4874
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.4137,
      "step": 4875
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.2311,
      "step": 4876
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.2776,
      "step": 4877
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.345,
      "step": 4878
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.9482,
      "step": 4879
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0002,
      "loss": 1.1436,
      "step": 4880
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.309,
      "step": 4881
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.955,
      "step": 4882
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 0.8769,
      "step": 4883
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.1543,
      "step": 4884
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.475,
      "step": 4885
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.2304,
      "step": 4886
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.8281,
      "step": 4887
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.2552,
      "step": 4888
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.1846,
      "step": 4889
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2039,
      "step": 4890
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.1789,
      "step": 4891
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0793,
      "step": 4892
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.0568,
      "step": 4893
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.1659,
      "step": 4894
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 0.6937,
      "step": 4895
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1206,
      "step": 4896
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.7179,
      "step": 4897
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.3989,
      "step": 4898
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.2101,
      "step": 4899
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.368,
      "step": 4900
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 0.8542,
      "step": 4901
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.2142,
      "step": 4902
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.2468,
      "step": 4903
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.9055,
      "step": 4904
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1411,
      "step": 4905
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.2848,
      "step": 4906
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.1266,
      "step": 4907
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.9473,
      "step": 4908
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.9791,
      "step": 4909
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.3426,
      "step": 4910
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.3478,
      "step": 4911
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.7847,
      "step": 4912
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.7104,
      "step": 4913
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.9879,
      "step": 4914
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.06,
      "step": 4915
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.7776,
      "step": 4916
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1037,
      "step": 4917
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.2488,
      "step": 4918
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 1.3676,
      "step": 4919
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.2127,
      "step": 4920
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.9242,
      "step": 4921
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.3286,
      "step": 4922
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.08251953125,
      "learning_rate": 0.0002,
      "loss": 0.4282,
      "step": 4923
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 0.8945,
      "step": 4924
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.8099,
      "step": 4925
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 0.4369,
      "step": 4926
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.9081,
      "step": 4927
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0453,
      "step": 4928
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.5238,
      "step": 4929
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.1236,
      "step": 4930
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.8023,
      "step": 4931
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.2977,
      "step": 4932
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0002,
      "loss": 1.1541,
      "step": 4933
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 0.8735,
      "step": 4934
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.0159,
      "step": 4935
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.9604,
      "step": 4936
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.9222,
      "step": 4937
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.1315,
      "step": 4938
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.9122,
      "step": 4939
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.3299,
      "step": 4940
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.014,
      "step": 4941
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.1262,
      "step": 4942
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 1.1516,
      "step": 4943
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.0311,
      "step": 4944
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 0.8509,
      "step": 4945
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.0543,
      "step": 4946
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.2331,
      "step": 4947
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.244,
      "step": 4948
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.2695,
      "step": 4949
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 1.601,
      "step": 4950
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.2097,
      "step": 4951
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.2106,
      "step": 4952
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.8673,
      "step": 4953
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.0334,
      "step": 4954
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.3299,
      "step": 4955
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.9196,
      "step": 4956
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2524,
      "step": 4957
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.8422,
      "step": 4958
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 1.067,
      "step": 4959
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.9313,
      "step": 4960
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 1.3304,
      "step": 4961
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.1713,
      "step": 4962
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.7789,
      "step": 4963
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.5868,
      "step": 4964
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0002,
      "loss": 1.0415,
      "step": 4965
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.0962,
      "step": 4966
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 0.8607,
      "step": 4967
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.3885,
      "step": 4968
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 0.8375,
      "step": 4969
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0002,
      "loss": 1.0805,
      "step": 4970
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.2086,
      "step": 4971
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.3487,
      "step": 4972
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0002,
      "loss": 0.5341,
      "step": 4973
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.0775,
      "step": 4974
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.0748,
      "step": 4975
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.1042,
      "step": 4976
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.2116,
      "step": 4977
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.1563,
      "step": 4978
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.381,
      "step": 4979
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.1997,
      "step": 4980
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2331,
      "step": 4981
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1867,
      "step": 4982
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0002,
      "loss": 0.5526,
      "step": 4983
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.431,
      "step": 4984
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.1016,
      "step": 4985
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 0.8636,
      "step": 4986
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.5895,
      "step": 4987
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.0745,
      "step": 4988
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 0.723,
      "step": 4989
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.1746,
      "step": 4990
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.0047,
      "step": 4991
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.0167,
      "step": 4992
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 0.7808,
      "step": 4993
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 0.5846,
      "step": 4994
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0002,
      "loss": 0.8829,
      "step": 4995
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.2343,
      "step": 4996
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.1071,
      "step": 4997
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 0.7865,
      "step": 4998
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.9018,
      "step": 4999
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.2226,
      "step": 5000
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0002,
      "loss": 1.018,
      "step": 5001
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0002,
      "loss": 1.011,
      "step": 5002
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.8969,
      "step": 5003
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0709,
      "step": 5004
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.3191,
      "step": 5005
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.3683,
      "step": 5006
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.0044,
      "step": 5007
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.8785,
      "step": 5008
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.1435,
      "step": 5009
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.1129,
      "step": 5010
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.1282,
      "step": 5011
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.5313,
      "step": 5012
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.3001,
      "step": 5013
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.1195,
      "step": 5014
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.3069,
      "step": 5015
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.2639,
      "step": 5016
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.3614,
      "step": 5017
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.1506,
      "step": 5018
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.0357,
      "step": 5019
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0002,
      "loss": 0.6775,
      "step": 5020
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.7963,
      "step": 5021
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0272,
      "step": 5022
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.1965,
      "step": 5023
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.3369,
      "step": 5024
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.04,
      "step": 5025
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 0.8504,
      "step": 5026
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.1298,
      "step": 5027
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 0.8779,
      "step": 5028
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.9318,
      "step": 5029
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.1361,
      "step": 5030
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.0864,
      "step": 5031
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0968,
      "step": 5032
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.1098,
      "step": 5033
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.0488,
      "step": 5034
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 1.2079,
      "step": 5035
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.2312,
      "step": 5036
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.22,
      "step": 5037
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 0.8604,
      "step": 5038
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.6188,
      "step": 5039
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.0393,
      "step": 5040
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.1754,
      "step": 5041
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.0214,
      "step": 5042
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.0437,
      "step": 5043
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.6309,
      "step": 5044
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 0.9072,
      "step": 5045
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.4341,
      "step": 5046
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.0541,
      "step": 5047
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 0.9108,
      "step": 5048
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.1384,
      "step": 5049
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.1051,
      "step": 5050
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 0.9913,
      "step": 5051
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.1818,
      "step": 5052
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.0422,
      "step": 5053
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.3179,
      "step": 5054
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.3616,
      "step": 5055
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0002,
      "loss": 0.5167,
      "step": 5056
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.229,
      "step": 5057
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.1687,
      "step": 5058
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.1937,
      "step": 5059
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1497,
      "step": 5060
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0051,
      "step": 5061
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.2059,
      "step": 5062
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.9929,
      "step": 5063
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.429,
      "step": 5064
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.1137,
      "step": 5065
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.1674,
      "step": 5066
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.2776,
      "step": 5067
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.7281,
      "step": 5068
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.0674,
      "step": 5069
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.0763,
      "step": 5070
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0002,
      "loss": 0.9318,
      "step": 5071
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.0934,
      "step": 5072
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 0.8521,
      "step": 5073
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.4415,
      "step": 5074
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.2679,
      "step": 5075
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.5903,
      "step": 5076
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.8065,
      "step": 5077
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.4752,
      "step": 5078
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.9963,
      "step": 5079
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.9842,
      "step": 5080
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.2587,
      "step": 5081
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0258,
      "step": 5082
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.998,
      "step": 5083
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.2547,
      "step": 5084
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.5786,
      "step": 5085
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.242,
      "step": 5086
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.1138,
      "step": 5087
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.3301,
      "step": 5088
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.0334,
      "step": 5089
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0002,
      "loss": 0.7281,
      "step": 5090
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0838,
      "step": 5091
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.6035,
      "step": 5092
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 0.9845,
      "step": 5093
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.4229,
      "step": 5094
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.4701,
      "step": 5095
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.6971,
      "step": 5096
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.0096,
      "step": 5097
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.1831,
      "step": 5098
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.1014,
      "step": 5099
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.9998,
      "step": 5100
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3975,
      "step": 5101
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.2367,
      "step": 5102
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.0701,
      "step": 5103
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 1.0234,
      "step": 5104
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.1844,
      "step": 5105
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.7219,
      "step": 5106
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0002,
      "loss": 1.0878,
      "step": 5107
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 0.845,
      "step": 5108
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.2743,
      "step": 5109
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.1946,
      "step": 5110
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.2344,
      "step": 5111
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.6706,
      "step": 5112
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 0.8825,
      "step": 5113
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.2895,
      "step": 5114
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.1524,
      "step": 5115
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.0194,
      "step": 5116
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.5159,
      "step": 5117
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.9293,
      "step": 5118
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.8169,
      "step": 5119
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.9564,
      "step": 5120
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.1523,
      "step": 5121
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.3875,
      "step": 5122
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.3414,
      "step": 5123
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.054,
      "step": 5124
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.8677,
      "step": 5125
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.2733,
      "step": 5126
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.1047,
      "step": 5127
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 0.8889,
      "step": 5128
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.8576,
      "step": 5129
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.5763,
      "step": 5130
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 0.8568,
      "step": 5131
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.0464,
      "step": 5132
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.7542,
      "step": 5133
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.4547,
      "step": 5134
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.0992,
      "step": 5135
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 0.9776,
      "step": 5136
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.5617,
      "step": 5137
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.0968,
      "step": 5138
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.004,
      "step": 5139
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.1372,
      "step": 5140
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.1399,
      "step": 5141
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.7133,
      "step": 5142
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.0835,
      "step": 5143
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.3262,
      "step": 5144
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2386,
      "step": 5145
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.8573,
      "step": 5146
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 0.8605,
      "step": 5147
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.2116,
      "step": 5148
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.2515,
      "step": 5149
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.2747,
      "step": 5150
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.4416,
      "step": 5151
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.0377,
      "step": 5152
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.0271,
      "step": 5153
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.2573,
      "step": 5154
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0563,
      "step": 5155
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.2611,
      "step": 5156
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 0.8237,
      "step": 5157
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.8857,
      "step": 5158
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 0.7978,
      "step": 5159
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.9575,
      "step": 5160
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.202,
      "step": 5161
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.0424,
      "step": 5162
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.3658,
      "step": 5163
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.28,
      "step": 5164
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 0.7828,
      "step": 5165
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.1559,
      "step": 5166
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.1604,
      "step": 5167
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 0.8438,
      "step": 5168
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 0.9607,
      "step": 5169
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0002,
      "loss": 0.7089,
      "step": 5170
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.277,
      "step": 5171
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.2466,
      "step": 5172
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.0921,
      "step": 5173
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.8165,
      "step": 5174
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.2793,
      "step": 5175
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.0869,
      "step": 5176
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.9925,
      "step": 5177
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2933,
      "step": 5178
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.1991,
      "step": 5179
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2317,
      "step": 5180
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.2494,
      "step": 5181
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.0839,
      "step": 5182
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.0367,
      "step": 5183
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.9207,
      "step": 5184
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.9509,
      "step": 5185
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.016,
      "step": 5186
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.7565,
      "step": 5187
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.1582,
      "step": 5188
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.2462,
      "step": 5189
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 0.9481,
      "step": 5190
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.0655,
      "step": 5191
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.1847,
      "step": 5192
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.0753,
      "step": 5193
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.0553,
      "step": 5194
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.475,
      "step": 5195
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.019,
      "step": 5196
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 0.8881,
      "step": 5197
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.0141,
      "step": 5198
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.0573,
      "step": 5199
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.05,
      "step": 5200
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 0.9044,
      "step": 5201
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 0.6401,
      "step": 5202
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.0579,
      "step": 5203
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.975,
      "step": 5204
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.3593,
      "step": 5205
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.9282,
      "step": 5206
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.0621,
      "step": 5207
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.0158,
      "step": 5208
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2948,
      "step": 5209
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 0.7767,
      "step": 5210
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.9987,
      "step": 5211
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.1873,
      "step": 5212
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.2568,
      "step": 5213
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3217,
      "step": 5214
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.2483,
      "step": 5215
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.0065,
      "step": 5216
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.2476,
      "step": 5217
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.8155,
      "step": 5218
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.8349,
      "step": 5219
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 0.8112,
      "step": 5220
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.1833,
      "step": 5221
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.1419,
      "step": 5222
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.2072,
      "step": 5223
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.09765625,
      "learning_rate": 0.0002,
      "loss": 0.5699,
      "step": 5224
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 0.9701,
      "step": 5225
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.3355,
      "step": 5226
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.2557,
      "step": 5227
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.2301,
      "step": 5228
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0002,
      "loss": 1.1274,
      "step": 5229
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.5154,
      "step": 5230
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0002,
      "loss": 0.9543,
      "step": 5231
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1106,
      "step": 5232
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.0393,
      "step": 5233
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.1193,
      "step": 5234
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1693,
      "step": 5235
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.2501,
      "step": 5236
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.3014,
      "step": 5237
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.0014,
      "step": 5238
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.1458,
      "step": 5239
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.2136,
      "step": 5240
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.3109,
      "step": 5241
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.398,
      "step": 5242
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.9785,
      "step": 5243
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.1603,
      "step": 5244
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.6351,
      "step": 5245
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0002,
      "loss": 0.9146,
      "step": 5246
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.3166,
      "step": 5247
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 0.931,
      "step": 5248
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.8551,
      "step": 5249
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.0862,
      "step": 5250
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.1174,
      "step": 5251
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.3388,
      "step": 5252
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.1985,
      "step": 5253
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.2422,
      "step": 5254
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.6788,
      "step": 5255
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.0269,
      "step": 5256
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.2375,
      "step": 5257
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.2989,
      "step": 5258
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 0.7921,
      "step": 5259
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 0.8974,
      "step": 5260
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.0784,
      "step": 5261
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.2191,
      "step": 5262
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 0.763,
      "step": 5263
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.6891,
      "step": 5264
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.0581,
      "step": 5265
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.9814,
      "step": 5266
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.1381,
      "step": 5267
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 0.758,
      "step": 5268
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 0.8979,
      "step": 5269
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.4184,
      "step": 5270
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.9571,
      "step": 5271
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 0.9284,
      "step": 5272
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 0.9103,
      "step": 5273
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.1004,
      "step": 5274
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.1131,
      "step": 5275
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.1811,
      "step": 5276
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.1249,
      "step": 5277
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.7734,
      "step": 5278
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.168,
      "step": 5279
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.0122,
      "step": 5280
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.4326,
      "step": 5281
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.0917,
      "step": 5282
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.2463,
      "step": 5283
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.0404,
      "step": 5284
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.8979,
      "step": 5285
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 0.9523,
      "step": 5286
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.8728,
      "step": 5287
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.2366,
      "step": 5288
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.466,
      "step": 5289
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.139,
      "step": 5290
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.9433,
      "step": 5291
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.0083,
      "step": 5292
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.3236,
      "step": 5293
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.3083,
      "step": 5294
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.1641,
      "step": 5295
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.1836,
      "step": 5296
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002,
      "loss": 1.4476,
      "step": 5297
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.0352,
      "step": 5298
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 0.7558,
      "step": 5299
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.0108,
      "step": 5300
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 0.3764,
      "step": 5301
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.9294,
      "step": 5302
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 0.8508,
      "step": 5303
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.183,
      "step": 5304
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.4338,
      "step": 5305
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.286,
      "step": 5306
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.1443,
      "step": 5307
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.1372,
      "step": 5308
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 0.8975,
      "step": 5309
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.2274,
      "step": 5310
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.1937,
      "step": 5311
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.3202,
      "step": 5312
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2439,
      "step": 5313
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.0909,
      "step": 5314
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.9315,
      "step": 5315
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 0.4411,
      "step": 5316
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.2659,
      "step": 5317
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.255,
      "step": 5318
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 0.8394,
      "step": 5319
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.0551,
      "step": 5320
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.8708,
      "step": 5321
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.2674,
      "step": 5322
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.1348,
      "step": 5323
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.2014,
      "step": 5324
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.2304,
      "step": 5325
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1982,
      "step": 5326
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.8071,
      "step": 5327
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.3874,
      "step": 5328
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.5565,
      "step": 5329
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.0331,
      "step": 5330
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.37,
      "step": 5331
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.1692,
      "step": 5332
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.2984,
      "step": 5333
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.168,
      "step": 5334
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.2605,
      "step": 5335
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.8366,
      "step": 5336
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.1632,
      "step": 5337
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.8968,
      "step": 5338
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 0.9893,
      "step": 5339
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.2427,
      "step": 5340
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1301,
      "step": 5341
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0002,
      "loss": 0.7426,
      "step": 5342
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.1969,
      "step": 5343
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1198,
      "step": 5344
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.9622,
      "step": 5345
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.5128,
      "step": 5346
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3098,
      "step": 5347
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.0091,
      "step": 5348
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.1411,
      "step": 5349
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0978,
      "step": 5350
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.4706,
      "step": 5351
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.4848,
      "step": 5352
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.0984,
      "step": 5353
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0638,
      "step": 5354
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.0774,
      "step": 5355
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.9417,
      "step": 5356
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 0.8637,
      "step": 5357
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.0167,
      "step": 5358
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0655,
      "step": 5359
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.2384,
      "step": 5360
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.1908,
      "step": 5361
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.1517,
      "step": 5362
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.3354,
      "step": 5363
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.1732,
      "step": 5364
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.5974,
      "step": 5365
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.1957,
      "step": 5366
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.1532,
      "step": 5367
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.3838,
      "step": 5368
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.2713,
      "step": 5369
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.2269,
      "step": 5370
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.7588,
      "step": 5371
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.8001,
      "step": 5372
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.5006,
      "step": 5373
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.4709,
      "step": 5374
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.1641,
      "step": 5375
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.1495,
      "step": 5376
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.1977,
      "step": 5377
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.0381,
      "step": 5378
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.158,
      "step": 5379
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.3611,
      "step": 5380
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.4452,
      "step": 5381
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.6028,
      "step": 5382
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.6063,
      "step": 5383
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 0.945,
      "step": 5384
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.06,
      "step": 5385
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.0858,
      "step": 5386
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.0573,
      "step": 5387
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.3754,
      "step": 5388
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.056,
      "step": 5389
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.0483,
      "step": 5390
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.5095,
      "step": 5391
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.3515,
      "step": 5392
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.8242,
      "step": 5393
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.1099,
      "step": 5394
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.109,
      "step": 5395
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.3227,
      "step": 5396
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.8957,
      "step": 5397
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 1.2202,
      "step": 5398
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.7316,
      "step": 5399
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3461,
      "step": 5400
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.0,
      "step": 5401
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.893,
      "step": 5402
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0002,
      "loss": 0.7453,
      "step": 5403
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.212,
      "step": 5404
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.0556,
      "step": 5405
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 0.9831,
      "step": 5406
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.0325,
      "step": 5407
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.445,
      "step": 5408
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.9615,
      "step": 5409
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 0.8632,
      "step": 5410
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.1765,
      "step": 5411
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.238,
      "step": 5412
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.4092,
      "step": 5413
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.7641,
      "step": 5414
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.9076,
      "step": 5415
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.436,
      "step": 5416
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.097,
      "step": 5417
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1474,
      "step": 5418
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 0.9609,
      "step": 5419
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.4265,
      "step": 5420
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.0744,
      "step": 5421
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.041,
      "step": 5422
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.1,
      "step": 5423
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.0942,
      "step": 5424
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 0.6745,
      "step": 5425
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.8765,
      "step": 5426
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0002,
      "loss": 0.9441,
      "step": 5427
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.3825,
      "step": 5428
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.3961,
      "step": 5429
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.0155,
      "step": 5430
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.1939,
      "step": 5431
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.0262,
      "step": 5432
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 0.8977,
      "step": 5433
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.6922,
      "step": 5434
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.8899,
      "step": 5435
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.1481,
      "step": 5436
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.957,
      "step": 5437
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.0718,
      "step": 5438
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0002,
      "loss": 1.0506,
      "step": 5439
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.2681,
      "step": 5440
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 1.3865,
      "step": 5441
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.4635,
      "step": 5442
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.7403,
      "step": 5443
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 1.3198,
      "step": 5444
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.005,
      "step": 5445
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002,
      "loss": 0.9576,
      "step": 5446
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.2747,
      "step": 5447
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.2455,
      "step": 5448
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.1275,
      "step": 5449
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.1227,
      "step": 5450
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.2767,
      "step": 5451
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.3002,
      "step": 5452
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.4283,
      "step": 5453
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.2107,
      "step": 5454
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.7038,
      "step": 5455
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.4201,
      "step": 5456
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 0.8831,
      "step": 5457
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.09228515625,
      "learning_rate": 0.0002,
      "loss": 0.8236,
      "step": 5458
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 0.9402,
      "step": 5459
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.3633,
      "step": 5460
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 0.7909,
      "step": 5461
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.2286,
      "step": 5462
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0002,
      "loss": 1.0163,
      "step": 5463
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.3823,
      "step": 5464
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 0.8143,
      "step": 5465
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.9171,
      "step": 5466
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.5088,
      "step": 5467
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.048,
      "step": 5468
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 0.9473,
      "step": 5469
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.1017,
      "step": 5470
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.8354,
      "step": 5471
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0492,
      "step": 5472
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.1994,
      "step": 5473
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2989,
      "step": 5474
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.2329,
      "step": 5475
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.5382,
      "step": 5476
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.6153,
      "step": 5477
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0002,
      "loss": 0.9345,
      "step": 5478
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.1246,
      "step": 5479
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1407,
      "step": 5480
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0002,
      "loss": 0.9281,
      "step": 5481
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.3452,
      "step": 5482
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 1.1735,
      "step": 5483
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.6423,
      "step": 5484
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.1121,
      "step": 5485
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0268,
      "step": 5486
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.6535,
      "step": 5487
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.1924,
      "step": 5488
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.1171,
      "step": 5489
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.1409,
      "step": 5490
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.2557,
      "step": 5491
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 0.7794,
      "step": 5492
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0002,
      "loss": 0.7519,
      "step": 5493
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 0.9562,
      "step": 5494
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 1.0152,
      "step": 5495
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.1157,
      "step": 5496
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.9655,
      "step": 5497
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0002,
      "loss": 0.4943,
      "step": 5498
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.1342,
      "step": 5499
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.1659,
      "step": 5500
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.1576,
      "step": 5501
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 0.8279,
      "step": 5502
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.0767,
      "step": 5503
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.9493,
      "step": 5504
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.318,
      "step": 5505
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.4012,
      "step": 5506
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0002,
      "loss": 0.5839,
      "step": 5507
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 0.8221,
      "step": 5508
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 0.7388,
      "step": 5509
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.1969,
      "step": 5510
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.057,
      "step": 5511
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.9832,
      "step": 5512
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.5259,
      "step": 5513
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 0.9343,
      "step": 5514
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.2431,
      "step": 5515
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.1682,
      "step": 5516
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.8457,
      "step": 5517
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 0.9367,
      "step": 5518
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 0.9506,
      "step": 5519
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.3792,
      "step": 5520
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.10400390625,
      "learning_rate": 0.0002,
      "loss": 0.8651,
      "step": 5521
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.6418,
      "step": 5522
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.7712,
      "step": 5523
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.217,
      "step": 5524
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.0211,
      "step": 5525
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3146,
      "step": 5526
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.1799,
      "step": 5527
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 1.2962,
      "step": 5528
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.2652,
      "step": 5529
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2828,
      "step": 5530
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.3525,
      "step": 5531
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.7567,
      "step": 5532
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.0538,
      "step": 5533
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.8728,
      "step": 5534
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.0335,
      "step": 5535
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 0.9235,
      "step": 5536
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.236,
      "step": 5537
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.2781,
      "step": 5538
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 0.8434,
      "step": 5539
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 0.9733,
      "step": 5540
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.7665,
      "step": 5541
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 0.6178,
      "step": 5542
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.2035,
      "step": 5543
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1725,
      "step": 5544
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 0.9413,
      "step": 5545
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.0145,
      "step": 5546
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.1628,
      "step": 5547
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1816,
      "step": 5548
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.4764,
      "step": 5549
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.0955,
      "step": 5550
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.8293,
      "step": 5551
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.1374,
      "step": 5552
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.4075,
      "step": 5553
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.1068,
      "step": 5554
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 0.9025,
      "step": 5555
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.9691,
      "step": 5556
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 0.9057,
      "step": 5557
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.6912,
      "step": 5558
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 0.6597,
      "step": 5559
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.348,
      "step": 5560
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.0859,
      "step": 5561
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.4191,
      "step": 5562
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 0.915,
      "step": 5563
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1741,
      "step": 5564
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 0.9448,
      "step": 5565
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 0.9377,
      "step": 5566
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.5699,
      "step": 5567
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 0.927,
      "step": 5568
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.2208,
      "step": 5569
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 0.7862,
      "step": 5570
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.2268,
      "step": 5571
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0002,
      "loss": 0.3099,
      "step": 5572
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.9924,
      "step": 5573
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.8654,
      "step": 5574
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.9598,
      "step": 5575
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.3498,
      "step": 5576
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.0749,
      "step": 5577
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.1216,
      "step": 5578
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.1973,
      "step": 5579
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.8459,
      "step": 5580
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.1565,
      "step": 5581
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.2258,
      "step": 5582
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.1078,
      "step": 5583
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 0.802,
      "step": 5584
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.2174,
      "step": 5585
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2549,
      "step": 5586
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 0.9245,
      "step": 5587
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 0.9055,
      "step": 5588
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.3474,
      "step": 5589
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 0.9727,
      "step": 5590
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.3016,
      "step": 5591
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.966,
      "step": 5592
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.2034,
      "step": 5593
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.7292,
      "step": 5594
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.9107,
      "step": 5595
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.8842,
      "step": 5596
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.5317,
      "step": 5597
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.9564,
      "step": 5598
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.8209,
      "step": 5599
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 0.8226,
      "step": 5600
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.0657,
      "step": 5601
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.3993,
      "step": 5602
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.1982,
      "step": 5603
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.3718,
      "step": 5604
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.1577,
      "step": 5605
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.2334,
      "step": 5606
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.1174,
      "step": 5607
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.1602,
      "step": 5608
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0352,
      "step": 5609
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.2859,
      "step": 5610
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 0.9068,
      "step": 5611
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3417,
      "step": 5612
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.617,
      "step": 5613
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.4554,
      "step": 5614
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.3045,
      "step": 5615
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.4319,
      "step": 5616
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.0174,
      "step": 5617
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 1.0549,
      "step": 5618
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.9172,
      "step": 5619
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.9807,
      "step": 5620
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.6928,
      "step": 5621
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.3944,
      "step": 5622
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.1139,
      "step": 5623
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.1961,
      "step": 5624
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.298,
      "step": 5625
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.9023,
      "step": 5626
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.1115,
      "step": 5627
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.2045,
      "step": 5628
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 0.9787,
      "step": 5629
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.1772,
      "step": 5630
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.0874,
      "step": 5631
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.1034,
      "step": 5632
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 0.8983,
      "step": 5633
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.6612,
      "step": 5634
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.7603,
      "step": 5635
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.1307,
      "step": 5636
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.293,
      "step": 5637
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.9479,
      "step": 5638
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.1023,
      "step": 5639
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.0284,
      "step": 5640
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.5477,
      "step": 5641
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.4633,
      "step": 5642
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.3417,
      "step": 5643
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.4255,
      "step": 5644
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.8578,
      "step": 5645
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.5619,
      "step": 5646
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 0.9782,
      "step": 5647
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.9137,
      "step": 5648
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.6477,
      "step": 5649
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.7537,
      "step": 5650
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.1295,
      "step": 5651
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 0.895,
      "step": 5652
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.9772,
      "step": 5653
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.0884,
      "step": 5654
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.8661,
      "step": 5655
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.9256,
      "step": 5656
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.024,
      "step": 5657
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 1.0948,
      "step": 5658
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 0.9021,
      "step": 5659
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 0.7996,
      "step": 5660
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.9518,
      "step": 5661
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.4076,
      "step": 5662
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0843,
      "step": 5663
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.259,
      "step": 5664
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.3422,
      "step": 5665
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.3898,
      "step": 5666
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.124,
      "step": 5667
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.9953,
      "step": 5668
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 0.7152,
      "step": 5669
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.0378,
      "step": 5670
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.4795,
      "step": 5671
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.1649,
      "step": 5672
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 0.9962,
      "step": 5673
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.1067,
      "step": 5674
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 0.8907,
      "step": 5675
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.2413,
      "step": 5676
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.0001,
      "step": 5677
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.3583,
      "step": 5678
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.1006,
      "step": 5679
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.1252,
      "step": 5680
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.2504,
      "step": 5681
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1209,
      "step": 5682
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 0.8202,
      "step": 5683
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.9913,
      "step": 5684
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.2368,
      "step": 5685
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 0.8351,
      "step": 5686
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1446,
      "step": 5687
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.1243,
      "step": 5688
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.3531,
      "step": 5689
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.0413,
      "step": 5690
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.2579,
      "step": 5691
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.2438,
      "step": 5692
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.2274,
      "step": 5693
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.0553,
      "step": 5694
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.8903,
      "step": 5695
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.4663,
      "step": 5696
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.1412,
      "step": 5697
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.0287,
      "step": 5698
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.3378,
      "step": 5699
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.1817,
      "step": 5700
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.0586,
      "step": 5701
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.8361,
      "step": 5702
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.7019,
      "step": 5703
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.0604,
      "step": 5704
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.2989,
      "step": 5705
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1118,
      "step": 5706
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.2912,
      "step": 5707
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.1703,
      "step": 5708
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.7789,
      "step": 5709
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0002,
      "loss": 1.3289,
      "step": 5710
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.9995,
      "step": 5711
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.8497,
      "step": 5712
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.2626,
      "step": 5713
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.139,
      "step": 5714
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.4215,
      "step": 5715
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.7478,
      "step": 5716
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.0375,
      "step": 5717
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 0.268,
      "step": 5718
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.2655,
      "step": 5719
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.1426,
      "step": 5720
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.821,
      "step": 5721
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.9533,
      "step": 5722
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.4625,
      "step": 5723
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.2206,
      "step": 5724
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.8025,
      "step": 5725
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.6593,
      "step": 5726
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.0086,
      "step": 5727
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 0.4853,
      "step": 5728
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.3386,
      "step": 5729
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 0.9644,
      "step": 5730
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.899,
      "step": 5731
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.0969,
      "step": 5732
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0707,
      "step": 5733
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0123,
      "step": 5734
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.1073,
      "step": 5735
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 0.8174,
      "step": 5736
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.9491,
      "step": 5737
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.168,
      "step": 5738
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.1568,
      "step": 5739
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.0252,
      "step": 5740
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.3649,
      "step": 5741
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.122,
      "step": 5742
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 0.8046,
      "step": 5743
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.1857,
      "step": 5744
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.5815,
      "step": 5745
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.4674,
      "step": 5746
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.8777,
      "step": 5747
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.1651,
      "step": 5748
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.8745,
      "step": 5749
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.3467,
      "step": 5750
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.3327,
      "step": 5751
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.1669,
      "step": 5752
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.7294,
      "step": 5753
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1697,
      "step": 5754
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.2247,
      "step": 5755
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.0736,
      "step": 5756
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.9923,
      "step": 5757
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.9423,
      "step": 5758
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.1862,
      "step": 5759
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.5002,
      "step": 5760
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.8385,
      "step": 5761
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.8395,
      "step": 5762
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.9843,
      "step": 5763
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.0498,
      "step": 5764
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0999,
      "step": 5765
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.3292,
      "step": 5766
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.1593,
      "step": 5767
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.7024,
      "step": 5768
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.1195,
      "step": 5769
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 0.9708,
      "step": 5770
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.0711,
      "step": 5771
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.1661,
      "step": 5772
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1966,
      "step": 5773
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 0.8499,
      "step": 5774
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.4919,
      "step": 5775
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2905,
      "step": 5776
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 0.9577,
      "step": 5777
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.9727,
      "step": 5778
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.0996,
      "step": 5779
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.505,
      "step": 5780
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.177,
      "step": 5781
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.0157,
      "step": 5782
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 0.996,
      "step": 5783
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.8753,
      "step": 5784
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.4596,
      "step": 5785
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.24,
      "step": 5786
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.8883,
      "step": 5787
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.3451,
      "step": 5788
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.1591,
      "step": 5789
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.092,
      "step": 5790
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.3566,
      "step": 5791
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.9973,
      "step": 5792
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.1343,
      "step": 5793
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.3005,
      "step": 5794
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.3802,
      "step": 5795
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.0084,
      "step": 5796
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 0.995,
      "step": 5797
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.0424,
      "step": 5798
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 0.9422,
      "step": 5799
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.4478,
      "step": 5800
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.1318,
      "step": 5801
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.9367,
      "step": 5802
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.06,
      "step": 5803
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.4276,
      "step": 5804
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1557,
      "step": 5805
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 0.953,
      "step": 5806
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.9977,
      "step": 5807
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.6215,
      "step": 5808
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.1866,
      "step": 5809
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.0905,
      "step": 5810
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1756,
      "step": 5811
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.4156,
      "step": 5812
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.0251,
      "step": 5813
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.7692,
      "step": 5814
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.9999,
      "step": 5815
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.7028,
      "step": 5816
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.1073,
      "step": 5817
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.871,
      "step": 5818
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.9073,
      "step": 5819
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.4868,
      "step": 5820
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.1172,
      "step": 5821
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 0.8782,
      "step": 5822
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.0617,
      "step": 5823
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.3466,
      "step": 5824
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.1659,
      "step": 5825
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.1437,
      "step": 5826
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.9191,
      "step": 5827
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 0.8162,
      "step": 5828
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0292,
      "step": 5829
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.0161,
      "step": 5830
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.4969,
      "step": 5831
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.2048,
      "step": 5832
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.4264,
      "step": 5833
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 0.6627,
      "step": 5834
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.9999,
      "step": 5835
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.0318,
      "step": 5836
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 1.1012,
      "step": 5837
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 0.9731,
      "step": 5838
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 0.7019,
      "step": 5839
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 0.9493,
      "step": 5840
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.1625,
      "step": 5841
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.1642,
      "step": 5842
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.9149,
      "step": 5843
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.945,
      "step": 5844
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.0607,
      "step": 5845
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.0118,
      "step": 5846
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.4954,
      "step": 5847
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.2164,
      "step": 5848
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.102,
      "step": 5849
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1069,
      "step": 5850
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.2693,
      "step": 5851
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.3908,
      "step": 5852
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 0.7822,
      "step": 5853
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.105,
      "step": 5854
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.025,
      "step": 5855
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 0.7435,
      "step": 5856
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0719,
      "step": 5857
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.3582,
      "step": 5858
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.6036,
      "step": 5859
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.1017,
      "step": 5860
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.5757,
      "step": 5861
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.1568,
      "step": 5862
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 0.8715,
      "step": 5863
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.1619,
      "step": 5864
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.2637,
      "step": 5865
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.1722,
      "step": 5866
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1629,
      "step": 5867
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.9625,
      "step": 5868
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 0.974,
      "step": 5869
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.9194,
      "step": 5870
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.4706,
      "step": 5871
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 1.0441,
      "step": 5872
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 0.9658,
      "step": 5873
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.3664,
      "step": 5874
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.0685,
      "step": 5875
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.229,
      "step": 5876
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.1741,
      "step": 5877
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.3198,
      "step": 5878
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.8937,
      "step": 5879
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.5277,
      "step": 5880
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0439,
      "step": 5881
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.9785,
      "step": 5882
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 1.2508,
      "step": 5883
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.2544,
      "step": 5884
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.9205,
      "step": 5885
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.098,
      "step": 5886
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.0371,
      "step": 5887
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 1.0181,
      "step": 5888
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.9439,
      "step": 5889
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 0.9674,
      "step": 5890
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.7359,
      "step": 5891
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.25,
      "step": 5892
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.4792,
      "step": 5893
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.8836,
      "step": 5894
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.091,
      "step": 5895
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.1329,
      "step": 5896
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.0435,
      "step": 5897
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 0.8932,
      "step": 5898
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002,
      "loss": 1.0433,
      "step": 5899
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.0734,
      "step": 5900
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.272,
      "step": 5901
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1216,
      "step": 5902
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.1978,
      "step": 5903
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.4074,
      "step": 5904
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.925,
      "step": 5905
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.1672,
      "step": 5906
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.1001,
      "step": 5907
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.3423,
      "step": 5908
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.8787,
      "step": 5909
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 0.9693,
      "step": 5910
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.7429,
      "step": 5911
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 0.8191,
      "step": 5912
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.4411,
      "step": 5913
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.2062,
      "step": 5914
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0666,
      "step": 5915
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.0083,
      "step": 5916
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1952,
      "step": 5917
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.1191,
      "step": 5918
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.7876,
      "step": 5919
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.1542,
      "step": 5920
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.2099,
      "step": 5921
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 0.9778,
      "step": 5922
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.9535,
      "step": 5923
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 0.884,
      "step": 5924
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.379,
      "step": 5925
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 0.9026,
      "step": 5926
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.2147,
      "step": 5927
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.0888671875,
      "learning_rate": 0.0002,
      "loss": 0.549,
      "step": 5928
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.4732,
      "step": 5929
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.0391,
      "step": 5930
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.2215,
      "step": 5931
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.2248,
      "step": 5932
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.1493,
      "step": 5933
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.3582,
      "step": 5934
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 0.7051,
      "step": 5935
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.9416,
      "step": 5936
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.0478,
      "step": 5937
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.1441,
      "step": 5938
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.9885,
      "step": 5939
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1563,
      "step": 5940
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.2509,
      "step": 5941
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.2053,
      "step": 5942
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.1208,
      "step": 5943
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0002,
      "loss": 1.1094,
      "step": 5944
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.0516,
      "step": 5945
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.2073,
      "step": 5946
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.9022,
      "step": 5947
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.4305,
      "step": 5948
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.7783,
      "step": 5949
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.1428,
      "step": 5950
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.4112,
      "step": 5951
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.8757,
      "step": 5952
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.9164,
      "step": 5953
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.9152,
      "step": 5954
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.4338,
      "step": 5955
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.1945,
      "step": 5956
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.5586,
      "step": 5957
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.3995,
      "step": 5958
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 0.9678,
      "step": 5959
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.1496,
      "step": 5960
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 0.978,
      "step": 5961
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 0.9313,
      "step": 5962
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.0696,
      "step": 5963
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.0717,
      "step": 5964
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0017,
      "step": 5965
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.5344,
      "step": 5966
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 0.8356,
      "step": 5967
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 0.8414,
      "step": 5968
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.0004,
      "step": 5969
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.0964,
      "step": 5970
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.1296,
      "step": 5971
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1102,
      "step": 5972
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0002,
      "loss": 0.661,
      "step": 5973
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.1325,
      "step": 5974
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.4206,
      "step": 5975
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.9738,
      "step": 5976
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2032,
      "step": 5977
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.0603,
      "step": 5978
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002,
      "loss": 1.3097,
      "step": 5979
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.4641,
      "step": 5980
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.474,
      "step": 5981
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.9091,
      "step": 5982
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.2161,
      "step": 5983
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2324,
      "step": 5984
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 1.0845,
      "step": 5985
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.4981,
      "step": 5986
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.9858,
      "step": 5987
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.5474,
      "step": 5988
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.034,
      "step": 5989
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.1613,
      "step": 5990
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.2562,
      "step": 5991
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.3935,
      "step": 5992
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.5296,
      "step": 5993
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 0.8665,
      "step": 5994
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1815,
      "step": 5995
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.1603,
      "step": 5996
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.0688,
      "step": 5997
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.0974,
      "step": 5998
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.3022,
      "step": 5999
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0902,
      "step": 6000
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.0932,
      "step": 6001
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 1.1297,
      "step": 6002
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.2903,
      "step": 6003
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.086,
      "step": 6004
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.9671,
      "step": 6005
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.2193,
      "step": 6006
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.347,
      "step": 6007
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.6714,
      "step": 6008
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.2765,
      "step": 6009
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 0.7741,
      "step": 6010
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.4736,
      "step": 6011
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0115,
      "step": 6012
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002,
      "loss": 1.1134,
      "step": 6013
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.0115,
      "step": 6014
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.1086,
      "step": 6015
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.0445,
      "step": 6016
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.9105,
      "step": 6017
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 0.8273,
      "step": 6018
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.1939,
      "step": 6019
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.2401,
      "step": 6020
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.973,
      "step": 6021
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.1771,
      "step": 6022
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0002,
      "loss": 0.9387,
      "step": 6023
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.0904,
      "step": 6024
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0002,
      "loss": 0.5374,
      "step": 6025
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.8368,
      "step": 6026
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.4512,
      "step": 6027
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.7741,
      "step": 6028
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 1.261,
      "step": 6029
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.2979,
      "step": 6030
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.9728,
      "step": 6031
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.119,
      "step": 6032
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.2551,
      "step": 6033
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 0.5534,
      "step": 6034
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.0114,
      "step": 6035
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.7694,
      "step": 6036
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.8182,
      "step": 6037
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.2122,
      "step": 6038
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.3671,
      "step": 6039
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.3162,
      "step": 6040
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.0401,
      "step": 6041
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.0742,
      "step": 6042
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.3319,
      "step": 6043
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.0747,
      "step": 6044
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.0435,
      "step": 6045
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.916,
      "step": 6046
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 0.9584,
      "step": 6047
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 0.9338,
      "step": 6048
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 0.92,
      "step": 6049
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.2871,
      "step": 6050
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.5676,
      "step": 6051
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.1371,
      "step": 6052
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.2077,
      "step": 6053
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1784,
      "step": 6054
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2754,
      "step": 6055
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.711,
      "step": 6056
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.5053,
      "step": 6057
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.9352,
      "step": 6058
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.8867,
      "step": 6059
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.3325,
      "step": 6060
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.9494,
      "step": 6061
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.2789,
      "step": 6062
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.2647,
      "step": 6063
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.8614,
      "step": 6064
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0583,
      "step": 6065
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.1402,
      "step": 6066
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.6117,
      "step": 6067
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.9806,
      "step": 6068
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.3542,
      "step": 6069
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1821,
      "step": 6070
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.156,
      "step": 6071
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1014,
      "step": 6072
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.0864,
      "step": 6073
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.1486,
      "step": 6074
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 0.9775,
      "step": 6075
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 1.1255,
      "step": 6076
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 0.8147,
      "step": 6077
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.0158,
      "step": 6078
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.2432,
      "step": 6079
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.364,
      "step": 6080
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.8423,
      "step": 6081
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 1.2173,
      "step": 6082
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.0793,
      "step": 6083
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 1.2526,
      "step": 6084
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.6076,
      "step": 6085
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0002,
      "loss": 0.9371,
      "step": 6086
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.0352,
      "step": 6087
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.1392,
      "step": 6088
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.97,
      "step": 6089
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.5827,
      "step": 6090
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.134,
      "step": 6091
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.9365,
      "step": 6092
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.8663,
      "step": 6093
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.2261,
      "step": 6094
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.9267,
      "step": 6095
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.9979,
      "step": 6096
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.8746,
      "step": 6097
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.9739,
      "step": 6098
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.8975,
      "step": 6099
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.8227,
      "step": 6100
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.3089,
      "step": 6101
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.367,
      "step": 6102
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.2046,
      "step": 6103
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.2407,
      "step": 6104
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.0194,
      "step": 6105
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.1431,
      "step": 6106
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 0.6602,
      "step": 6107
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0002,
      "loss": 0.8402,
      "step": 6108
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.0182,
      "step": 6109
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.5125,
      "step": 6110
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.9214,
      "step": 6111
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 0.957,
      "step": 6112
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.6248,
      "step": 6113
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.0279,
      "step": 6114
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.2679,
      "step": 6115
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.2236,
      "step": 6116
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.9824,
      "step": 6117
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.7616,
      "step": 6118
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.0439,
      "step": 6119
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2493,
      "step": 6120
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002,
      "loss": 1.2077,
      "step": 6121
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.0409,
      "step": 6122
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.0114,
      "step": 6123
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.6072,
      "step": 6124
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.2139,
      "step": 6125
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.0531,
      "step": 6126
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2681,
      "step": 6127
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.0035,
      "step": 6128
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.0668,
      "step": 6129
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.3413,
      "step": 6130
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.168,
      "step": 6131
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.1994,
      "step": 6132
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2133,
      "step": 6133
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.0954,
      "step": 6134
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.3845,
      "step": 6135
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.9679,
      "step": 6136
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.2858,
      "step": 6137
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.7175,
      "step": 6138
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.9414,
      "step": 6139
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 0.78,
      "step": 6140
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.9193,
      "step": 6141
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.9685,
      "step": 6142
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.0089,
      "step": 6143
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.4217,
      "step": 6144
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 1.3629,
      "step": 6145
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.3979,
      "step": 6146
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.9414,
      "step": 6147
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 1.2371,
      "step": 6148
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.3696,
      "step": 6149
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 0.9778,
      "step": 6150
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.2429,
      "step": 6151
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.9001,
      "step": 6152
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.9493,
      "step": 6153
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.8404,
      "step": 6154
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.1707,
      "step": 6155
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.7204,
      "step": 6156
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 0.9609,
      "step": 6157
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.0917,
      "step": 6158
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.1944,
      "step": 6159
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.1568,
      "step": 6160
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.2178,
      "step": 6161
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.1138,
      "step": 6162
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.8013,
      "step": 6163
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 0.9901,
      "step": 6164
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.1681,
      "step": 6165
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.3816,
      "step": 6166
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.1859,
      "step": 6167
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 1.0359,
      "step": 6168
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.1766,
      "step": 6169
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.6745,
      "step": 6170
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.8421,
      "step": 6171
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.3449,
      "step": 6172
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 0.9177,
      "step": 6173
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.1439,
      "step": 6174
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.9919,
      "step": 6175
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.7924,
      "step": 6176
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 0.5583,
      "step": 6177
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.942,
      "step": 6178
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.4168,
      "step": 6179
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.4909,
      "step": 6180
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.1895,
      "step": 6181
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.8858,
      "step": 6182
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 0.8387,
      "step": 6183
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.1107,
      "step": 6184
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.1578,
      "step": 6185
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.8596,
      "step": 6186
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.1972,
      "step": 6187
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.4068,
      "step": 6188
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.8704,
      "step": 6189
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0654,
      "step": 6190
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.8452,
      "step": 6191
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.1391,
      "step": 6192
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.9057,
      "step": 6193
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.3886,
      "step": 6194
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.3461,
      "step": 6195
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0157,
      "step": 6196
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1282,
      "step": 6197
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.0416,
      "step": 6198
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.103,
      "step": 6199
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.236,
      "step": 6200
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.05,
      "step": 6201
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.2494,
      "step": 6202
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.221,
      "step": 6203
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.9538,
      "step": 6204
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.1025,
      "step": 6205
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.1419,
      "step": 6206
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.3632,
      "step": 6207
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.2195,
      "step": 6208
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0556,
      "step": 6209
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 0.9814,
      "step": 6210
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3662,
      "step": 6211
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.1677,
      "step": 6212
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.9829,
      "step": 6213
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.1343,
      "step": 6214
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.3546,
      "step": 6215
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.7759,
      "step": 6216
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.2063,
      "step": 6217
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 0.9903,
      "step": 6218
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.0141,
      "step": 6219
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.4416,
      "step": 6220
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2386,
      "step": 6221
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.1356,
      "step": 6222
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.437,
      "step": 6223
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 0.8579,
      "step": 6224
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 0.876,
      "step": 6225
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.3346,
      "step": 6226
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.069,
      "step": 6227
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.1707,
      "step": 6228
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.0985,
      "step": 6229
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0613,
      "step": 6230
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.0738,
      "step": 6231
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.9922,
      "step": 6232
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.8322,
      "step": 6233
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.1219,
      "step": 6234
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.1734,
      "step": 6235
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 0.8146,
      "step": 6236
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.1051,
      "step": 6237
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.2364,
      "step": 6238
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.3442,
      "step": 6239
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.09,
      "step": 6240
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 0.9193,
      "step": 6241
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 0.8516,
      "step": 6242
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.3113,
      "step": 6243
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 0.966,
      "step": 6244
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.0036,
      "step": 6245
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.4691,
      "step": 6246
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.0847,
      "step": 6247
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 0.5801,
      "step": 6248
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.2603,
      "step": 6249
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.3627,
      "step": 6250
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.0303,
      "step": 6251
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.2396,
      "step": 6252
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.0539,
      "step": 6253
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 0.6986,
      "step": 6254
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 0.9044,
      "step": 6255
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.8382,
      "step": 6256
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.1407,
      "step": 6257
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.0063,
      "step": 6258
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.235,
      "step": 6259
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.9288,
      "step": 6260
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.8677,
      "step": 6261
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.0533,
      "step": 6262
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 0.9288,
      "step": 6263
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.1506,
      "step": 6264
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.2661,
      "step": 6265
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 0.5985,
      "step": 6266
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.7735,
      "step": 6267
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.5471,
      "step": 6268
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.2523,
      "step": 6269
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.3741,
      "step": 6270
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 0.3987,
      "step": 6271
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.2668,
      "step": 6272
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 0.7625,
      "step": 6273
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.5997,
      "step": 6274
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.0655,
      "step": 6275
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 0.7778,
      "step": 6276
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 0.9451,
      "step": 6277
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.0241,
      "step": 6278
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.3353,
      "step": 6279
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 0.8847,
      "step": 6280
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.3258,
      "step": 6281
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.4417,
      "step": 6282
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.2722,
      "step": 6283
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0136,
      "step": 6284
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.0097,
      "step": 6285
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3905,
      "step": 6286
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1907,
      "step": 6287
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.2845,
      "step": 6288
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.882,
      "step": 6289
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.9029,
      "step": 6290
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.0218,
      "step": 6291
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.2542,
      "step": 6292
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.8981,
      "step": 6293
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.9865,
      "step": 6294
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.1594,
      "step": 6295
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 0.9388,
      "step": 6296
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.0106,
      "step": 6297
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.5966,
      "step": 6298
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 0.9344,
      "step": 6299
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0002,
      "loss": 0.4354,
      "step": 6300
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.0989,
      "step": 6301
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 1.1279,
      "step": 6302
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 0.974,
      "step": 6303
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.9576,
      "step": 6304
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.1964,
      "step": 6305
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 1.1942,
      "step": 6306
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.5184,
      "step": 6307
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 0.9188,
      "step": 6308
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.1163,
      "step": 6309
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2282,
      "step": 6310
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.9499,
      "step": 6311
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.9249,
      "step": 6312
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.1531,
      "step": 6313
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.3355,
      "step": 6314
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.2917,
      "step": 6315
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 0.9141,
      "step": 6316
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.8901,
      "step": 6317
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.1676,
      "step": 6318
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 0.8675,
      "step": 6319
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.0515,
      "step": 6320
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.2133,
      "step": 6321
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.7005,
      "step": 6322
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.9475,
      "step": 6323
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.8764,
      "step": 6324
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.1625,
      "step": 6325
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1261,
      "step": 6326
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.6407,
      "step": 6327
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0541,
      "step": 6328
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.9091,
      "step": 6329
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.1174,
      "step": 6330
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.2892,
      "step": 6331
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.1912,
      "step": 6332
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.1217,
      "step": 6333
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0002,
      "loss": 1.0081,
      "step": 6334
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.7329,
      "step": 6335
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.4127,
      "step": 6336
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.0914,
      "step": 6337
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.4566,
      "step": 6338
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.1848,
      "step": 6339
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.1393,
      "step": 6340
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.192,
      "step": 6341
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.9641,
      "step": 6342
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.1676,
      "step": 6343
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.3499,
      "step": 6344
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.416,
      "step": 6345
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.3443,
      "step": 6346
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.232,
      "step": 6347
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 0.6306,
      "step": 6348
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.8524,
      "step": 6349
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.0618,
      "step": 6350
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.0373,
      "step": 6351
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 0.7278,
      "step": 6352
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.9724,
      "step": 6353
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.9487,
      "step": 6354
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.934,
      "step": 6355
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0491,
      "step": 6356
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.9943,
      "step": 6357
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 0.968,
      "step": 6358
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.0055,
      "step": 6359
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 0.7857,
      "step": 6360
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.5299,
      "step": 6361
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.744,
      "step": 6362
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.2394,
      "step": 6363
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1206,
      "step": 6364
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 0.8078,
      "step": 6365
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 0.9399,
      "step": 6366
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.9752,
      "step": 6367
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.1464,
      "step": 6368
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.2476,
      "step": 6369
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 0.9685,
      "step": 6370
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.122,
      "step": 6371
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.7776,
      "step": 6372
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.9237,
      "step": 6373
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.9761,
      "step": 6374
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 0.8975,
      "step": 6375
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0246,
      "step": 6376
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.037,
      "step": 6377
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 0.927,
      "step": 6378
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.0606,
      "step": 6379
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.4116,
      "step": 6380
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.6101,
      "step": 6381
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 0.9352,
      "step": 6382
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 1.0471,
      "step": 6383
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.0451,
      "step": 6384
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.1246,
      "step": 6385
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.0919,
      "step": 6386
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.1288,
      "step": 6387
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.9652,
      "step": 6388
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.1492,
      "step": 6389
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.0938,
      "step": 6390
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 0.9444,
      "step": 6391
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.9159,
      "step": 6392
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.8792,
      "step": 6393
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.5405,
      "step": 6394
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.0792,
      "step": 6395
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.16,
      "step": 6396
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1667,
      "step": 6397
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0264,
      "step": 6398
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.992,
      "step": 6399
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.7947,
      "step": 6400
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.1137,
      "step": 6401
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.1046,
      "step": 6402
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.2132,
      "step": 6403
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.4216,
      "step": 6404
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.4246,
      "step": 6405
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.3275,
      "step": 6406
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.0773,
      "step": 6407
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.1509,
      "step": 6408
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.9187,
      "step": 6409
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.4169,
      "step": 6410
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0002,
      "loss": 0.8299,
      "step": 6411
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.9149,
      "step": 6412
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.9564,
      "step": 6413
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.0577,
      "step": 6414
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.3686,
      "step": 6415
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.132,
      "step": 6416
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.8719,
      "step": 6417
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0298,
      "step": 6418
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.2113,
      "step": 6419
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.3281,
      "step": 6420
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.8966,
      "step": 6421
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.2823,
      "step": 6422
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.0721,
      "step": 6423
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.2924,
      "step": 6424
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.8915,
      "step": 6425
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.3615,
      "step": 6426
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.3208,
      "step": 6427
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 0.2959,
      "step": 6428
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.2758,
      "step": 6429
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.0377,
      "step": 6430
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.0414,
      "step": 6431
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 0.6863,
      "step": 6432
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.1983,
      "step": 6433
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.2943,
      "step": 6434
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 0.9419,
      "step": 6435
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.0996,
      "step": 6436
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 0.8021,
      "step": 6437
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.133,
      "step": 6438
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.2309,
      "step": 6439
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.1052,
      "step": 6440
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.1167,
      "step": 6441
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.5276,
      "step": 6442
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.0384,
      "step": 6443
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0002,
      "loss": 0.876,
      "step": 6444
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.929,
      "step": 6445
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.0886,
      "step": 6446
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.0719,
      "step": 6447
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002,
      "loss": 0.2723,
      "step": 6448
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1169,
      "step": 6449
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.2094,
      "step": 6450
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.9529,
      "step": 6451
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.5244,
      "step": 6452
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.0042,
      "step": 6453
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 0.8513,
      "step": 6454
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 0.764,
      "step": 6455
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 0.776,
      "step": 6456
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.2436,
      "step": 6457
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.0515,
      "step": 6458
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.2558,
      "step": 6459
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.107,
      "step": 6460
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.0999,
      "step": 6461
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.127,
      "step": 6462
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.107,
      "step": 6463
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.1215,
      "step": 6464
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.4329,
      "step": 6465
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.2057,
      "step": 6466
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.0245,
      "step": 6467
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.09912109375,
      "learning_rate": 0.0002,
      "loss": 0.5252,
      "step": 6468
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 0.8539,
      "step": 6469
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.8013,
      "step": 6470
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.9761,
      "step": 6471
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.7614,
      "step": 6472
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.2048,
      "step": 6473
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.8562,
      "step": 6474
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.9194,
      "step": 6475
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.4221,
      "step": 6476
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.21,
      "step": 6477
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.2659,
      "step": 6478
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 0.5699,
      "step": 6479
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.1912,
      "step": 6480
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 1.4082,
      "step": 6481
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.9224,
      "step": 6482
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0878,
      "step": 6483
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.6651,
      "step": 6484
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.123,
      "step": 6485
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.995,
      "step": 6486
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.1972,
      "step": 6487
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.3285,
      "step": 6488
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 0.9073,
      "step": 6489
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.2099,
      "step": 6490
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.9478,
      "step": 6491
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 0.9889,
      "step": 6492
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.8477,
      "step": 6493
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.8467,
      "step": 6494
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.2015,
      "step": 6495
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 0.7628,
      "step": 6496
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.009,
      "step": 6497
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.9485,
      "step": 6498
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.36,
      "step": 6499
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1558,
      "step": 6500
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.9847,
      "step": 6501
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 0.857,
      "step": 6502
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.0314,
      "step": 6503
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0002,
      "loss": 0.7183,
      "step": 6504
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.3802,
      "step": 6505
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 1.379,
      "step": 6506
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1875,
      "learning_rate": 0.0002,
      "loss": 1.1984,
      "step": 6507
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 0.7727,
      "step": 6508
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.2782,
      "step": 6509
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.9459,
      "step": 6510
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.1865,
      "step": 6511
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.9592,
      "step": 6512
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.0332,
      "step": 6513
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.0004,
      "step": 6514
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.2289,
      "step": 6515
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.2685,
      "step": 6516
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.0848,
      "step": 6517
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.3508,
      "step": 6518
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.2678,
      "step": 6519
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.2244,
      "step": 6520
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.2306,
      "step": 6521
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.0743,
      "step": 6522
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.3136,
      "step": 6523
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.8353,
      "step": 6524
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.1526,
      "step": 6525
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.849,
      "step": 6526
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 0.6361,
      "step": 6527
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.1818,
      "step": 6528
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0886,
      "step": 6529
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 0.9329,
      "step": 6530
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.6973,
      "step": 6531
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.0893,
      "step": 6532
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.4077,
      "step": 6533
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.9969,
      "step": 6534
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.9567,
      "step": 6535
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.0995,
      "step": 6536
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.0394,
      "step": 6537
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 1.0619,
      "step": 6538
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.2219,
      "step": 6539
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0572,
      "step": 6540
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.1706,
      "step": 6541
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002,
      "loss": 1.4867,
      "step": 6542
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.7652,
      "step": 6543
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.4262,
      "step": 6544
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 0.7805,
      "step": 6545
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.2918,
      "step": 6546
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.2875,
      "step": 6547
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.3283,
      "step": 6548
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.2726,
      "step": 6549
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.0693,
      "step": 6550
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.9671,
      "step": 6551
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.0901,
      "step": 6552
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0031,
      "step": 6553
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 0.2808,
      "step": 6554
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.1122,
      "step": 6555
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.2054,
      "step": 6556
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.8112,
      "step": 6557
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.1671,
      "step": 6558
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 0.7303,
      "step": 6559
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.0014,
      "step": 6560
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.0249,
      "step": 6561
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.1734,
      "step": 6562
    },
    {
      "epoch": 1.08,
      "grad_norm": 2.625,
      "learning_rate": 0.0002,
      "loss": 1.753,
      "step": 6563
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.8337,
      "step": 6564
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.8389,
      "step": 6565
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1614,
      "step": 6566
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 0.6579,
      "step": 6567
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 1.1542,
      "step": 6568
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.2733,
      "step": 6569
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0002,
      "loss": 1.0325,
      "step": 6570
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.185,
      "step": 6571
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.0278,
      "step": 6572
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.0588,
      "step": 6573
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 0.2178,
      "step": 6574
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.21,
      "step": 6575
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.8181,
      "step": 6576
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.6071,
      "step": 6577
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.1352,
      "step": 6578
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.2513,
      "step": 6579
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.2966,
      "step": 6580
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.2414,
      "step": 6581
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 0.9689,
      "step": 6582
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.8038,
      "step": 6583
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.0604,
      "step": 6584
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.413,
      "step": 6585
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.1051,
      "step": 6586
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.1212,
      "step": 6587
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.8082,
      "step": 6588
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.8948,
      "step": 6589
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.2503,
      "step": 6590
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.731,
      "step": 6591
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.1368,
      "step": 6592
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.9422,
      "step": 6593
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 0.8984,
      "step": 6594
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1032,
      "step": 6595
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.1182,
      "step": 6596
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.4886,
      "step": 6597
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.1611,
      "step": 6598
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.2462,
      "step": 6599
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 0.9634,
      "step": 6600
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.8415,
      "step": 6601
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.9779,
      "step": 6602
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.9213,
      "step": 6603
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.0964,
      "step": 6604
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0298,
      "step": 6605
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.2363,
      "step": 6606
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.0787,
      "step": 6607
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.9396,
      "step": 6608
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.2806,
      "step": 6609
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3297,
      "step": 6610
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.7437,
      "step": 6611
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.1777,
      "step": 6612
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.8645,
      "step": 6613
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.1153,
      "step": 6614
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.2743,
      "step": 6615
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 0.902,
      "step": 6616
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.5888,
      "step": 6617
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 0.7264,
      "step": 6618
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.1273,
      "step": 6619
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 0.9766,
      "step": 6620
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.3965,
      "step": 6621
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.222,
      "step": 6622
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.8833,
      "step": 6623
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.0795,
      "step": 6624
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 0.64,
      "step": 6625
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.8215,
      "step": 6626
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.3818,
      "step": 6627
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 1.7749,
      "step": 6628
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 0.8762,
      "step": 6629
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.9178,
      "step": 6630
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.5192,
      "step": 6631
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.1309,
      "step": 6632
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.0152,
      "step": 6633
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3465,
      "step": 6634
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.0249,
      "step": 6635
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0002,
      "loss": 0.7099,
      "step": 6636
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.6147,
      "step": 6637
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.1589,
      "step": 6638
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.3349,
      "step": 6639
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.0505,
      "step": 6640
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0699,
      "step": 6641
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.1297,
      "step": 6642
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.0755,
      "step": 6643
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.1448,
      "step": 6644
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.1303,
      "step": 6645
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.478,
      "step": 6646
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.6541,
      "step": 6647
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.3078,
      "step": 6648
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.1257,
      "step": 6649
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 0.8605,
      "step": 6650
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 0.983,
      "step": 6651
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.1001,
      "step": 6652
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.1798,
      "step": 6653
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.2886,
      "step": 6654
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.0339,
      "step": 6655
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.151,
      "step": 6656
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.9889,
      "step": 6657
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0421,
      "step": 6658
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.9893,
      "step": 6659
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.9437,
      "step": 6660
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.054,
      "step": 6661
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.8917,
      "step": 6662
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.922,
      "step": 6663
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.2065,
      "step": 6664
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 0.9652,
      "step": 6665
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.5073,
      "step": 6666
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.4868,
      "step": 6667
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 1.0117,
      "step": 6668
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.055,
      "step": 6669
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.5026,
      "step": 6670
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.89,
      "step": 6671
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.293,
      "step": 6672
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.036,
      "step": 6673
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0002,
      "loss": 0.5533,
      "step": 6674
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.008,
      "step": 6675
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.1771,
      "step": 6676
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.8518,
      "step": 6677
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.7484,
      "step": 6678
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.6835,
      "step": 6679
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 0.9721,
      "step": 6680
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.9946,
      "step": 6681
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 0.9692,
      "step": 6682
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 0.7474,
      "step": 6683
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.7046,
      "step": 6684
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.2371,
      "step": 6685
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.1243,
      "step": 6686
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.8701,
      "step": 6687
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.064,
      "step": 6688
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.2115,
      "step": 6689
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 0.9435,
      "step": 6690
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.2718,
      "step": 6691
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.2446,
      "step": 6692
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.0773,
      "step": 6693
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.2207,
      "step": 6694
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.8794,
      "step": 6695
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.2168,
      "step": 6696
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.2722,
      "step": 6697
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.1011,
      "step": 6698
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 0.8507,
      "step": 6699
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.9442,
      "step": 6700
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.2477,
      "step": 6701
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.2193,
      "step": 6702
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 0.7707,
      "step": 6703
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.0732,
      "step": 6704
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.3114,
      "step": 6705
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.2053,
      "step": 6706
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.3369,
      "step": 6707
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.1299,
      "step": 6708
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.1858,
      "step": 6709
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 0.9983,
      "step": 6710
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.3715,
      "step": 6711
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.1203,
      "step": 6712
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0002,
      "loss": 0.8849,
      "step": 6713
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 0.6274,
      "step": 6714
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.814,
      "step": 6715
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 0.9527,
      "step": 6716
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.9185,
      "step": 6717
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.7209,
      "step": 6718
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.3239,
      "step": 6719
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.2717,
      "step": 6720
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.0516,
      "step": 6721
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.4705,
      "step": 6722
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.2788,
      "step": 6723
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.2831,
      "step": 6724
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.3589,
      "step": 6725
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0267,
      "step": 6726
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.1506,
      "step": 6727
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.3315,
      "step": 6728
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.0651,
      "step": 6729
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.407,
      "step": 6730
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0002,
      "loss": 0.4908,
      "step": 6731
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.2284,
      "step": 6732
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.9863,
      "step": 6733
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 0.5755,
      "step": 6734
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.2051,
      "step": 6735
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.0345,
      "step": 6736
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.9655,
      "step": 6737
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0607,
      "step": 6738
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.0731,
      "step": 6739
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.0451,
      "step": 6740
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.9284,
      "step": 6741
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.9249,
      "step": 6742
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.0052,
      "step": 6743
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.007,
      "step": 6744
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.3105,
      "step": 6745
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.9036,
      "step": 6746
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.7955,
      "step": 6747
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.0446,
      "step": 6748
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.8676,
      "step": 6749
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1044,
      "step": 6750
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.1072,
      "step": 6751
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.7836,
      "step": 6752
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.1768,
      "step": 6753
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.9601,
      "step": 6754
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0342,
      "step": 6755
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.1213,
      "step": 6756
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 0.8107,
      "step": 6757
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.7852,
      "step": 6758
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.4795,
      "step": 6759
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 1.1157,
      "step": 6760
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 1.3362,
      "step": 6761
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.9088,
      "step": 6762
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.0533,
      "step": 6763
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 0.8642,
      "step": 6764
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0002,
      "loss": 1.1071,
      "step": 6765
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.6611,
      "step": 6766
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.1945,
      "step": 6767
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.9008,
      "step": 6768
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.1438,
      "step": 6769
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0002,
      "loss": 0.3729,
      "step": 6770
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.2991,
      "step": 6771
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.0105,
      "step": 6772
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 0.8995,
      "step": 6773
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.1512,
      "step": 6774
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.912,
      "step": 6775
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.9876,
      "step": 6776
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.2282,
      "step": 6777
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.025,
      "step": 6778
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.1343,
      "step": 6779
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.8597,
      "step": 6780
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.3068,
      "step": 6781
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 0.8908,
      "step": 6782
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.8565,
      "step": 6783
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.1666,
      "step": 6784
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.0656,
      "step": 6785
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.9802,
      "step": 6786
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 0.8812,
      "step": 6787
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.8829,
      "step": 6788
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.9858,
      "step": 6789
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.9656,
      "step": 6790
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.385,
      "step": 6791
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0002,
      "loss": 0.705,
      "step": 6792
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.2013,
      "step": 6793
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.9108,
      "step": 6794
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0002,
      "loss": 0.7123,
      "step": 6795
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.0152,
      "step": 6796
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.93,
      "step": 6797
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.7452,
      "step": 6798
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 0.8159,
      "step": 6799
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.1449,
      "step": 6800
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 0.7841,
      "step": 6801
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.701,
      "step": 6802
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.0465,
      "step": 6803
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.2741,
      "step": 6804
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 0.697,
      "step": 6805
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 0.9992,
      "step": 6806
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.2407,
      "step": 6807
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 0.8547,
      "step": 6808
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.0256,
      "step": 6809
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 0.9428,
      "step": 6810
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.2892,
      "step": 6811
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.2127,
      "step": 6812
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.886,
      "step": 6813
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 0.8774,
      "step": 6814
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.5525,
      "step": 6815
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.0221,
      "step": 6816
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.328,
      "step": 6817
    },
    {
      "epoch": 1.12,
      "grad_norm": 2.734375,
      "learning_rate": 0.0002,
      "loss": 0.9492,
      "step": 6818
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.3539,
      "step": 6819
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.1586,
      "step": 6820
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.9356,
      "step": 6821
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.3335,
      "step": 6822
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.7675,
      "step": 6823
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.0728,
      "step": 6824
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.9087,
      "step": 6825
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.551,
      "step": 6826
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.2159,
      "step": 6827
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.9304,
      "step": 6828
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.3701,
      "step": 6829
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.9694,
      "step": 6830
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 1.0204,
      "step": 6831
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.806,
      "step": 6832
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 0.825,
      "step": 6833
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.3818,
      "step": 6834
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.1656,
      "step": 6835
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.0797,
      "step": 6836
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.3216,
      "step": 6837
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 0.9662,
      "step": 6838
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.3889,
      "step": 6839
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.132,
      "step": 6840
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 0.8896,
      "step": 6841
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.7526,
      "step": 6842
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 0.9797,
      "step": 6843
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.0158,
      "step": 6844
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.65625,
      "learning_rate": 0.0002,
      "loss": 0.7579,
      "step": 6845
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 0.9772,
      "step": 6846
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.6616,
      "step": 6847
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.1212,
      "step": 6848
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.9886,
      "step": 6849
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.3016,
      "step": 6850
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.2941,
      "step": 6851
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 0.7796,
      "step": 6852
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 0.6441,
      "step": 6853
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.782,
      "step": 6854
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002,
      "loss": 1.2776,
      "step": 6855
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.4256,
      "step": 6856
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.2947,
      "step": 6857
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.016,
      "step": 6858
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 0.8734,
      "step": 6859
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 1.1059,
      "step": 6860
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 0.8241,
      "step": 6861
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 0.9512,
      "step": 6862
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.664,
      "step": 6863
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.2205,
      "step": 6864
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.27,
      "step": 6865
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.8583,
      "step": 6866
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002,
      "loss": 1.2867,
      "step": 6867
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.6893,
      "step": 6868
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.1706,
      "step": 6869
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.0498,
      "step": 6870
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 0.8958,
      "step": 6871
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.3634,
      "step": 6872
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.9295,
      "step": 6873
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.7679,
      "step": 6874
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.7664,
      "step": 6875
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.299,
      "step": 6876
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.653,
      "step": 6877
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.2715,
      "step": 6878
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.0349,
      "step": 6879
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.6181,
      "step": 6880
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 0.9774,
      "step": 6881
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.3217,
      "step": 6882
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.5886,
      "step": 6883
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0953,
      "step": 6884
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.097,
      "step": 6885
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 0.8544,
      "step": 6886
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.0831,
      "step": 6887
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.3608,
      "step": 6888
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.7326,
      "step": 6889
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.0608,
      "step": 6890
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.9122,
      "step": 6891
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 1.0258,
      "step": 6892
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.788,
      "step": 6893
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.9368,
      "step": 6894
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.9968,
      "step": 6895
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.0935,
      "step": 6896
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 0.4467,
      "step": 6897
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.7837,
      "step": 6898
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.1973,
      "step": 6899
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 0.9615,
      "step": 6900
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.0911,
      "step": 6901
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0002,
      "loss": 0.8797,
      "step": 6902
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 0.9585,
      "step": 6903
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.1482,
      "step": 6904
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.0321,
      "step": 6905
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 0.9974,
      "step": 6906
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.0045,
      "step": 6907
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 0.8976,
      "step": 6908
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.0752,
      "step": 6909
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0318,
      "step": 6910
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 0.7941,
      "step": 6911
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.1173,
      "step": 6912
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.3529,
      "step": 6913
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.3554,
      "step": 6914
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.7493,
      "step": 6915
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0291,
      "step": 6916
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.0594,
      "step": 6917
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.2547,
      "step": 6918
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.0735,
      "step": 6919
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 0.8499,
      "step": 6920
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.2082,
      "step": 6921
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.2209,
      "step": 6922
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.334,
      "step": 6923
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.1734,
      "step": 6924
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.013,
      "step": 6925
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.1104,
      "step": 6926
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.08935546875,
      "learning_rate": 0.0002,
      "loss": 0.4331,
      "step": 6927
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0002,
      "loss": 0.4852,
      "step": 6928
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.1555,
      "step": 6929
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.7267,
      "step": 6930
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.898,
      "step": 6931
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.4216,
      "step": 6932
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1445,
      "step": 6933
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0008,
      "step": 6934
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.3242,
      "step": 6935
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.1858,
      "step": 6936
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.1381,
      "step": 6937
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.9035,
      "step": 6938
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.3481,
      "step": 6939
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2378,
      "step": 6940
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.2063,
      "step": 6941
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 0.9553,
      "step": 6942
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.9045,
      "step": 6943
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 0.9104,
      "step": 6944
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.2,
      "step": 6945
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.6213,
      "step": 6946
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.2218,
      "step": 6947
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.2452,
      "step": 6948
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1098,
      "step": 6949
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.8808,
      "step": 6950
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.0453,
      "step": 6951
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.2332,
      "step": 6952
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.8395,
      "step": 6953
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.181,
      "step": 6954
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 0.9415,
      "step": 6955
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.2257,
      "step": 6956
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.1396,
      "step": 6957
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.1511,
      "step": 6958
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 1.1131,
      "step": 6959
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 0.931,
      "step": 6960
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.9209,
      "step": 6961
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0346,
      "step": 6962
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 0.8298,
      "step": 6963
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.2267,
      "step": 6964
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.0712,
      "step": 6965
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.1674,
      "step": 6966
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.1772,
      "step": 6967
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.9298,
      "step": 6968
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.7268,
      "step": 6969
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 1.2288,
      "step": 6970
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.0702,
      "step": 6971
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.4028,
      "step": 6972
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 0.9259,
      "step": 6973
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 0.9803,
      "step": 6974
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.3412,
      "step": 6975
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.2477,
      "step": 6976
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 1.1458,
      "step": 6977
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.0872,
      "step": 6978
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.2916,
      "step": 6979
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.8091,
      "step": 6980
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.7521,
      "step": 6981
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002,
      "loss": 1.179,
      "step": 6982
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0002,
      "loss": 0.7181,
      "step": 6983
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 0.8487,
      "step": 6984
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.2841,
      "step": 6985
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.4959,
      "step": 6986
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1044,
      "step": 6987
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.0095,
      "step": 6988
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 0.91,
      "step": 6989
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.1714,
      "step": 6990
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.4286,
      "step": 6991
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.0042,
      "step": 6992
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.472,
      "step": 6993
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.9083,
      "step": 6994
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 0.9859,
      "step": 6995
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002,
      "loss": 1.4237,
      "step": 6996
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.0433,
      "step": 6997
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.8517,
      "step": 6998
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0967,
      "step": 6999
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.1969,
      "step": 7000
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.9009,
      "step": 7001
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 0.7715,
      "step": 7002
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 0.9745,
      "step": 7003
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.8236,
      "step": 7004
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.7565,
      "step": 7005
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.9441,
      "step": 7006
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.4772,
      "step": 7007
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0049,
      "step": 7008
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.0233,
      "step": 7009
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 0.7307,
      "step": 7010
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.8593,
      "step": 7011
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.9625,
      "step": 7012
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.7834,
      "step": 7013
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.4558,
      "step": 7014
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.1141,
      "step": 7015
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 0.9153,
      "step": 7016
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.0829,
      "step": 7017
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.2998,
      "step": 7018
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.3218,
      "step": 7019
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.244,
      "step": 7020
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.1771,
      "step": 7021
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.4519,
      "step": 7022
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.1096,
      "step": 7023
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 0.8493,
      "step": 7024
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.0647,
      "step": 7025
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.2248,
      "step": 7026
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.9657,
      "step": 7027
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.0527,
      "step": 7028
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 0.9683,
      "step": 7029
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.8709,
      "step": 7030
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 0.7653,
      "step": 7031
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.382,
      "step": 7032
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 0.9957,
      "step": 7033
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002,
      "loss": 1.3391,
      "step": 7034
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.0388,
      "step": 7035
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.9616,
      "step": 7036
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1739,
      "step": 7037
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.1128,
      "step": 7038
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1428,
      "step": 7039
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.3452,
      "step": 7040
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.5277,
      "step": 7041
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.8708,
      "step": 7042
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.0667,
      "step": 7043
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 0.2716,
      "step": 7044
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.6199,
      "step": 7045
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 0.7464,
      "step": 7046
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.147,
      "step": 7047
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.2555,
      "step": 7048
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 0.8127,
      "step": 7049
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.1869,
      "step": 7050
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.2721,
      "step": 7051
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2095,
      "step": 7052
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0002,
      "loss": 0.9445,
      "step": 7053
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.102,
      "step": 7054
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.0251,
      "step": 7055
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.2767,
      "step": 7056
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.8073,
      "step": 7057
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.958,
      "step": 7058
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.0972,
      "step": 7059
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.1746,
      "step": 7060
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.1152,
      "step": 7061
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.6573,
      "step": 7062
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.8047,
      "step": 7063
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.2242,
      "step": 7064
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.2146,
      "step": 7065
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.053,
      "step": 7066
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.1038,
      "step": 7067
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.0262,
      "step": 7068
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.4824,
      "step": 7069
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.4222,
      "step": 7070
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 1.0961,
      "step": 7071
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.6832,
      "step": 7072
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.4853,
      "step": 7073
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.145,
      "step": 7074
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 0.8284,
      "step": 7075
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.0268,
      "step": 7076
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.3006,
      "step": 7077
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.3568,
      "step": 7078
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.1527,
      "step": 7079
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.8334,
      "step": 7080
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.9991,
      "step": 7081
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002,
      "loss": 1.1339,
      "step": 7082
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.9915,
      "step": 7083
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.1939,
      "step": 7084
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.8843,
      "step": 7085
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.3019,
      "step": 7086
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.1578,
      "step": 7087
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 0.9322,
      "step": 7088
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 0.3273,
      "step": 7089
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.9111,
      "step": 7090
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.7339,
      "step": 7091
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.4027,
      "step": 7092
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.5773,
      "step": 7093
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.1792,
      "step": 7094
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.5937,
      "step": 7095
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.9957,
      "step": 7096
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.3709,
      "step": 7097
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.4585,
      "step": 7098
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.8181,
      "step": 7099
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.2576,
      "step": 7100
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.8823,
      "step": 7101
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.2702,
      "step": 7102
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.1064,
      "step": 7103
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.1916,
      "step": 7104
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 0.8087,
      "step": 7105
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.1173,
      "step": 7106
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0425,
      "step": 7107
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.8682,
      "step": 7108
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 0.8269,
      "step": 7109
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.1182,
      "step": 7110
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.208,
      "step": 7111
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 0.8926,
      "step": 7112
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0002,
      "loss": 0.7231,
      "step": 7113
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.9155,
      "step": 7114
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.3692,
      "step": 7115
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.2615,
      "step": 7116
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002,
      "loss": 0.3913,
      "step": 7117
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 0.9638,
      "step": 7118
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.0931,
      "step": 7119
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 0.9023,
      "step": 7120
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 0.8671,
      "step": 7121
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.1893,
      "step": 7122
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 1.0476,
      "step": 7123
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.3111,
      "step": 7124
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.1464,
      "step": 7125
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.923,
      "step": 7126
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.9099,
      "step": 7127
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 0.8555,
      "step": 7128
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.0561,
      "step": 7129
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.8291,
      "step": 7130
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.0101,
      "step": 7131
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.0381,
      "step": 7132
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.9122,
      "step": 7133
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 0.923,
      "step": 7134
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 1.19,
      "step": 7135
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.9383,
      "step": 7136
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 0.9558,
      "step": 7137
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.9688,
      "step": 7138
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 1.1934,
      "step": 7139
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.0253,
      "step": 7140
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.3633,
      "step": 7141
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.9193,
      "step": 7142
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.2068,
      "step": 7143
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.422,
      "step": 7144
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.7394,
      "step": 7145
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.9947,
      "step": 7146
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.2693,
      "step": 7147
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.9516,
      "step": 7148
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.9411,
      "step": 7149
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.1437,
      "step": 7150
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.62,
      "step": 7151
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.3585,
      "step": 7152
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.9023,
      "step": 7153
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.1787,
      "step": 7154
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 0.8412,
      "step": 7155
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.5229,
      "step": 7156
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.8658,
      "step": 7157
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.0139,
      "step": 7158
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 0.9773,
      "step": 7159
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.2396,
      "step": 7160
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 0.7867,
      "step": 7161
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.9158,
      "step": 7162
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.5013,
      "step": 7163
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.0625,
      "step": 7164
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.7279,
      "step": 7165
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.6858,
      "step": 7166
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.9761,
      "step": 7167
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 0.9063,
      "step": 7168
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.1707,
      "step": 7169
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.9596,
      "step": 7170
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.9594,
      "step": 7171
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.2377,
      "step": 7172
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 0.868,
      "step": 7173
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.1661,
      "step": 7174
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.2875,
      "step": 7175
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.063,
      "step": 7176
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.0558,
      "step": 7177
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.831,
      "step": 7178
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.4183,
      "step": 7179
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.3482,
      "step": 7180
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.8915,
      "step": 7181
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.9652,
      "step": 7182
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002,
      "loss": 0.9319,
      "step": 7183
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.1072,
      "step": 7184
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.017,
      "step": 7185
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.1729,
      "step": 7186
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 0.7293,
      "step": 7187
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.0441,
      "step": 7188
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002,
      "loss": 1.4149,
      "step": 7189
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.2437,
      "step": 7190
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.0884,
      "step": 7191
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.8651,
      "step": 7192
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.9111,
      "step": 7193
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.369,
      "step": 7194
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.1294,
      "step": 7195
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.1619,
      "step": 7196
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.5673,
      "step": 7197
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.8184,
      "step": 7198
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.4502,
      "step": 7199
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.3268,
      "step": 7200
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.9712,
      "step": 7201
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.3013,
      "step": 7202
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.1537,
      "step": 7203
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.6356,
      "step": 7204
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.9093,
      "step": 7205
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.1279,
      "step": 7206
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 0.985,
      "step": 7207
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.24,
      "step": 7208
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.8892,
      "step": 7209
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.4659,
      "step": 7210
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 0.8509,
      "step": 7211
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.7113,
      "step": 7212
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.0539,
      "step": 7213
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.0982,
      "step": 7214
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.0003,
      "step": 7215
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.464,
      "step": 7216
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.081,
      "step": 7217
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 0.9109,
      "step": 7218
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.8513,
      "step": 7219
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.084,
      "step": 7220
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.3322,
      "step": 7221
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 0.7966,
      "step": 7222
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.9803,
      "step": 7223
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 0.988,
      "step": 7224
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.0103,
      "step": 7225
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.5866,
      "step": 7226
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.2006,
      "step": 7227
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.7446,
      "step": 7228
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.6418,
      "step": 7229
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.3407,
      "step": 7230
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.2659,
      "step": 7231
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.9249,
      "step": 7232
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 0.914,
      "step": 7233
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.0218,
      "step": 7234
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.2364,
      "step": 7235
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.2537,
      "step": 7236
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.2691,
      "step": 7237
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.0569,
      "step": 7238
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.3036,
      "step": 7239
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.3212,
      "step": 7240
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 1.2332,
      "step": 7241
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0681,
      "step": 7242
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.4063,
      "step": 7243
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.632,
      "step": 7244
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.7453,
      "step": 7245
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.068,
      "step": 7246
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.12,
      "step": 7247
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.2653,
      "step": 7248
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.557,
      "step": 7249
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.1875,
      "step": 7250
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.0931,
      "step": 7251
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.0987,
      "step": 7252
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.212,
      "step": 7253
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.2887,
      "step": 7254
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0002,
      "loss": 0.4073,
      "step": 7255
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 0.8119,
      "step": 7256
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.2357,
      "step": 7257
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.1996,
      "step": 7258
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.0379,
      "step": 7259
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 0.408,
      "step": 7260
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.9254,
      "step": 7261
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.281,
      "step": 7262
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 0.6581,
      "step": 7263
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.3179,
      "step": 7264
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.2126,
      "step": 7265
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.7244,
      "step": 7266
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 0.9594,
      "step": 7267
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.2051,
      "step": 7268
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.5918,
      "step": 7269
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 0.9827,
      "step": 7270
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 0.7751,
      "step": 7271
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.2297,
      "step": 7272
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.3319,
      "step": 7273
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.3141,
      "step": 7274
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.7767,
      "step": 7275
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.2954,
      "step": 7276
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.1333,
      "step": 7277
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 0.2531,
      "step": 7278
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.7769,
      "step": 7279
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.2596,
      "step": 7280
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.1784,
      "step": 7281
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0002,
      "loss": 1.5003,
      "step": 7282
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 0.7866,
      "step": 7283
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.3041,
      "step": 7284
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.3854,
      "step": 7285
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.3094,
      "step": 7286
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.0085,
      "step": 7287
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.4129,
      "step": 7288
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 0.7838,
      "step": 7289
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.2735,
      "step": 7290
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 0.9236,
      "step": 7291
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.1743,
      "step": 7292
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 0.9876,
      "step": 7293
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.1572,
      "step": 7294
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 0.6145,
      "step": 7295
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.1391,
      "step": 7296
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 0.9513,
      "step": 7297
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.5624,
      "step": 7298
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1158,
      "step": 7299
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.0046,
      "step": 7300
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.1782,
      "step": 7301
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.6487,
      "step": 7302
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.0368,
      "step": 7303
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.003,
      "step": 7304
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.0152,
      "step": 7305
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.2516,
      "step": 7306
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.4051,
      "step": 7307
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.1533,
      "step": 7308
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.4312,
      "step": 7309
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.2287,
      "step": 7310
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 1.1144,
      "step": 7311
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.043,
      "step": 7312
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.7714,
      "step": 7313
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1875,
      "learning_rate": 0.0002,
      "loss": 1.254,
      "step": 7314
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.0414,
      "step": 7315
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.1691,
      "step": 7316
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.1242,
      "step": 7317
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.0015,
      "step": 7318
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 0.9898,
      "step": 7319
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.1113,
      "step": 7320
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 1.1165,
      "step": 7321
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 0.9455,
      "step": 7322
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.8756,
      "step": 7323
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 0.8343,
      "step": 7324
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0002,
      "loss": 0.9,
      "step": 7325
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.8281,
      "step": 7326
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 0.7886,
      "step": 7327
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.3292,
      "step": 7328
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.8857,
      "step": 7329
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.0986,
      "step": 7330
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.9662,
      "step": 7331
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.9508,
      "step": 7332
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.369,
      "step": 7333
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.198,
      "step": 7334
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.2882,
      "step": 7335
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.027,
      "step": 7336
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 0.964,
      "step": 7337
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 0.8942,
      "step": 7338
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.2835,
      "step": 7339
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 0.9099,
      "step": 7340
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.1077,
      "step": 7341
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.153,
      "step": 7342
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 1.449,
      "step": 7343
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.0398,
      "step": 7344
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.1259,
      "step": 7345
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.0326,
      "step": 7346
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.545,
      "step": 7347
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 0.9705,
      "step": 7348
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 0.8008,
      "step": 7349
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 0.9864,
      "step": 7350
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.2554,
      "step": 7351
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 0.8958,
      "step": 7352
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.2756,
      "step": 7353
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 0.8754,
      "step": 7354
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 0.9299,
      "step": 7355
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.9427,
      "step": 7356
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.1088,
      "step": 7357
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 0.9187,
      "step": 7358
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 0.6591,
      "step": 7359
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.2023,
      "step": 7360
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.1596,
      "step": 7361
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.0495,
      "step": 7362
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.218,
      "step": 7363
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.2709,
      "step": 7364
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.0594,
      "step": 7365
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.2474,
      "step": 7366
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 0.4325,
      "step": 7367
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 0.9613,
      "step": 7368
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.0328,
      "step": 7369
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.1545,
      "step": 7370
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.8632,
      "step": 7371
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.0721,
      "step": 7372
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 0.5315,
      "step": 7373
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.2473,
      "step": 7374
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.0793,
      "step": 7375
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 1.3177,
      "step": 7376
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0092,
      "step": 7377
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.8569,
      "step": 7378
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.6724,
      "step": 7379
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0002,
      "loss": 1.0123,
      "step": 7380
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.1639,
      "step": 7381
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.9211,
      "step": 7382
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 1.225,
      "step": 7383
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.0333,
      "step": 7384
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 0.9667,
      "step": 7385
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 0.9259,
      "step": 7386
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.3683,
      "step": 7387
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.5222,
      "step": 7388
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.4085,
      "step": 7389
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.2191,
      "step": 7390
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.8204,
      "step": 7391
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.0109,
      "step": 7392
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.3873,
      "step": 7393
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.1541,
      "step": 7394
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.776,
      "step": 7395
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.1058,
      "step": 7396
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.1899,
      "step": 7397
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 0.9121,
      "step": 7398
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.2394,
      "step": 7399
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 0.8854,
      "step": 7400
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.5435,
      "step": 7401
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.1481,
      "step": 7402
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.0471,
      "step": 7403
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002,
      "loss": 0.8348,
      "step": 7404
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 0.8453,
      "step": 7405
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.6361,
      "step": 7406
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.1236,
      "step": 7407
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.2822,
      "step": 7408
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.0551,
      "step": 7409
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 0.9369,
      "step": 7410
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.8292,
      "step": 7411
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 1.2659,
      "step": 7412
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 0.9799,
      "step": 7413
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.1229,
      "step": 7414
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0002,
      "loss": 0.5106,
      "step": 7415
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.0039,
      "step": 7416
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.9103,
      "step": 7417
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.0366,
      "step": 7418
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 0.9762,
      "step": 7419
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0002,
      "loss": 0.6609,
      "step": 7420
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1875,
      "learning_rate": 0.0002,
      "loss": 0.8099,
      "step": 7421
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.2138,
      "step": 7422
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.216,
      "step": 7423
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.6119,
      "step": 7424
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 0.8188,
      "step": 7425
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 0.9905,
      "step": 7426
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.2506,
      "step": 7427
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.0519,
      "step": 7428
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.4581,
      "step": 7429
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1637,
      "step": 7430
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.3842,
      "step": 7431
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.0005,
      "step": 7432
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 0.8178,
      "step": 7433
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.019,
      "step": 7434
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.1108,
      "step": 7435
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 0.8144,
      "step": 7436
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.1242,
      "step": 7437
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.9934,
      "step": 7438
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 0.8024,
      "step": 7439
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 0.7767,
      "step": 7440
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.0973,
      "step": 7441
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.1486,
      "step": 7442
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.9736,
      "step": 7443
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.9836,
      "step": 7444
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.2478,
      "step": 7445
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 0.8737,
      "step": 7446
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.9735,
      "step": 7447
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.9719,
      "step": 7448
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 1.0569,
      "step": 7449
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.6988,
      "step": 7450
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.2871,
      "step": 7451
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.86,
      "step": 7452
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.2008,
      "step": 7453
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 0.8868,
      "step": 7454
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.1602,
      "step": 7455
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.1407,
      "step": 7456
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.1965,
      "step": 7457
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.2956,
      "step": 7458
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.0349,
      "step": 7459
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.0908203125,
      "learning_rate": 0.0002,
      "loss": 0.164,
      "step": 7460
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.3549,
      "step": 7461
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.8727,
      "step": 7462
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0002,
      "loss": 1.2687,
      "step": 7463
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1875,
      "learning_rate": 0.0002,
      "loss": 1.1663,
      "step": 7464
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.3162,
      "step": 7465
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 0.9389,
      "step": 7466
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.3673,
      "step": 7467
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 0.9083,
      "step": 7468
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.2433,
      "step": 7469
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 0.9767,
      "step": 7470
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.0016,
      "step": 7471
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0002,
      "loss": 0.844,
      "step": 7472
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.9359,
      "step": 7473
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.319,
      "step": 7474
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.7021,
      "step": 7475
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.9294,
      "step": 7476
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.0568,
      "step": 7477
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.6665,
      "step": 7478
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.3353,
      "step": 7479
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.2291,
      "step": 7480
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.0023,
      "step": 7481
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.0328,
      "step": 7482
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 0.7877,
      "step": 7483
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.0152,
      "step": 7484
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 0.9843,
      "step": 7485
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.0016,
      "step": 7486
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.335,
      "step": 7487
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.0054,
      "step": 7488
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.1599,
      "step": 7489
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.9908,
      "step": 7490
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0002,
      "loss": 0.8112,
      "step": 7491
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.3888,
      "step": 7492
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 0.8544,
      "step": 7493
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 0.8057,
      "step": 7494
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002,
      "loss": 1.422,
      "step": 7495
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 1.0333,
      "step": 7496
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.8,
      "step": 7497
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002,
      "loss": 1.2764,
      "step": 7498
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.1417,
      "step": 7499
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 0.8984,
      "step": 7500
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.8211,
      "step": 7501
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.2081,
      "step": 7502
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.1482,
      "step": 7503
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.2103,
      "step": 7504
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.8843,
      "step": 7505
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.2157,
      "step": 7506
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.5322,
      "step": 7507
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.2541,
      "step": 7508
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.109375,
      "learning_rate": 0.0002,
      "loss": 0.6977,
      "step": 7509
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0002,
      "loss": 0.9232,
      "step": 7510
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.8738,
      "step": 7511
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.9509,
      "step": 7512
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.0594,
      "step": 7513
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.9878,
      "step": 7514
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.759,
      "step": 7515
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.3259,
      "step": 7516
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.5863,
      "step": 7517
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.041,
      "step": 7518
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.1761,
      "step": 7519
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.9546,
      "step": 7520
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.2119,
      "step": 7521
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.0293,
      "step": 7522
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.0807,
      "step": 7523
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.754,
      "step": 7524
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.3665,
      "step": 7525
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 0.914,
      "step": 7526
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.7343,
      "step": 7527
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.9963,
      "step": 7528
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 0.9549,
      "step": 7529
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.9676,
      "step": 7530
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.9822,
      "step": 7531
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.1355,
      "step": 7532
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.6164,
      "step": 7533
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.0452,
      "step": 7534
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.7511,
      "step": 7535
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 0.4267,
      "step": 7536
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.1769,
      "step": 7537
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.0053,
      "step": 7538
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.1599,
      "step": 7539
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.576,
      "step": 7540
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.1294,
      "step": 7541
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.8497,
      "step": 7542
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.2092,
      "step": 7543
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.9625,
      "step": 7544
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.2163,
      "step": 7545
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 0.8972,
      "step": 7546
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.933,
      "step": 7547
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.0351,
      "step": 7548
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.7513,
      "step": 7549
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.0756,
      "step": 7550
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.1173,
      "step": 7551
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.5047,
      "step": 7552
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.421,
      "step": 7553
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.1728,
      "step": 7554
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.2071,
      "step": 7555
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.9462,
      "step": 7556
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.1044,
      "step": 7557
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.0347,
      "step": 7558
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.2164,
      "step": 7559
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.09619140625,
      "learning_rate": 0.0002,
      "loss": 0.4438,
      "step": 7560
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0002,
      "loss": 0.4488,
      "step": 7561
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.982,
      "step": 7562
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.3706,
      "step": 7563
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.2581,
      "step": 7564
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.9784,
      "step": 7565
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.0179,
      "step": 7566
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.0338,
      "step": 7567
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.9186,
      "step": 7568
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 0.9923,
      "step": 7569
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.4152,
      "step": 7570
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 0.3856,
      "step": 7571
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 0.802,
      "step": 7572
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 0.8519,
      "step": 7573
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.1243,
      "step": 7574
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.9081,
      "step": 7575
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.8532,
      "step": 7576
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.2683,
      "step": 7577
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.0271,
      "step": 7578
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.4049,
      "step": 7579
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.1953,
      "step": 7580
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.3922,
      "step": 7581
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 0.5506,
      "step": 7582
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 0.9654,
      "step": 7583
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0002,
      "loss": 1.5151,
      "step": 7584
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.2723,
      "step": 7585
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 0.7902,
      "step": 7586
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.3313,
      "step": 7587
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.0493,
      "step": 7588
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.022,
      "step": 7589
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.1315,
      "step": 7590
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.8511,
      "step": 7591
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.0949,
      "step": 7592
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 0.726,
      "step": 7593
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3142,
      "step": 7594
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 0.9158,
      "step": 7595
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.055,
      "step": 7596
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.2334,
      "step": 7597
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 0.9306,
      "step": 7598
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.258,
      "step": 7599
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2244,
      "step": 7600
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.0516,
      "step": 7601
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 0.8945,
      "step": 7602
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.1086,
      "step": 7603
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.1392,
      "step": 7604
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.867,
      "step": 7605
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.1232,
      "step": 7606
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.3952,
      "step": 7607
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.1407,
      "step": 7608
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 0.8174,
      "step": 7609
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.9706,
      "step": 7610
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.6436,
      "step": 7611
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.1052,
      "step": 7612
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.0617,
      "step": 7613
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 0.9497,
      "step": 7614
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.0018,
      "step": 7615
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.0754,
      "step": 7616
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.3062,
      "step": 7617
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.9093,
      "step": 7618
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.0572,
      "step": 7619
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.09423828125,
      "learning_rate": 0.0002,
      "loss": 0.5244,
      "step": 7620
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.0959,
      "step": 7621
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.2443,
      "step": 7622
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.1676,
      "step": 7623
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 0.7978,
      "step": 7624
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 0.8628,
      "step": 7625
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.2762,
      "step": 7626
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.2437,
      "step": 7627
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.1081,
      "step": 7628
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.1356,
      "step": 7629
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 1.3619,
      "step": 7630
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.6701,
      "step": 7631
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.8901,
      "step": 7632
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.0728,
      "step": 7633
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 0.9672,
      "step": 7634
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.0436,
      "step": 7635
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.3768,
      "step": 7636
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.1235,
      "step": 7637
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.828,
      "step": 7638
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.2167,
      "step": 7639
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.2186,
      "step": 7640
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.0683,
      "step": 7641
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.0891,
      "step": 7642
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 0.9037,
      "step": 7643
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 0.833,
      "step": 7644
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 0.8556,
      "step": 7645
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.4285,
      "step": 7646
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.2755,
      "step": 7647
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.4114,
      "step": 7648
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.2452,
      "step": 7649
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.057,
      "step": 7650
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.512,
      "step": 7651
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.825,
      "step": 7652
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.9553,
      "step": 7653
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.1375,
      "step": 7654
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 0.8792,
      "step": 7655
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.0515,
      "step": 7656
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.8194,
      "step": 7657
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.6442,
      "step": 7658
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.1095,
      "step": 7659
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.0144,
      "step": 7660
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 1.1806,
      "step": 7661
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.3005,
      "step": 7662
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.4955,
      "step": 7663
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.0582,
      "step": 7664
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.0554,
      "step": 7665
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.7422,
      "step": 7666
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 0.9466,
      "step": 7667
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.2727,
      "step": 7668
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.21,
      "step": 7669
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.8937,
      "step": 7670
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.3044,
      "step": 7671
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.1984,
      "step": 7672
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.376,
      "step": 7673
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 0.8481,
      "step": 7674
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.4971,
      "step": 7675
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.0237,
      "step": 7676
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.9447,
      "step": 7677
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.1529,
      "step": 7678
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.5655,
      "step": 7679
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.0998,
      "step": 7680
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 0.9679,
      "step": 7681
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.9834,
      "step": 7682
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.0526,
      "step": 7683
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 1.0234,
      "step": 7684
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.278,
      "step": 7685
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.9276,
      "step": 7686
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.0961,
      "step": 7687
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.0975,
      "step": 7688
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.1793,
      "step": 7689
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.074,
      "step": 7690
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.2475,
      "step": 7691
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.9743,
      "step": 7692
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.2553,
      "step": 7693
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.7991,
      "step": 7694
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0171,
      "step": 7695
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.4003,
      "step": 7696
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.1992,
      "step": 7697
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.1,
      "step": 7698
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.1389,
      "step": 7699
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002,
      "loss": 1.2178,
      "step": 7700
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 0.9913,
      "step": 7701
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 0.8105,
      "step": 7702
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.9376,
      "step": 7703
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.4751,
      "step": 7704
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.4086,
      "step": 7705
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.4956,
      "step": 7706
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 0.644,
      "step": 7707
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.896,
      "step": 7708
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.1077,
      "step": 7709
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.1475,
      "step": 7710
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3001,
      "step": 7711
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.009,
      "step": 7712
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.2131,
      "step": 7713
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.0238,
      "step": 7714
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.8328,
      "step": 7715
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.4021,
      "step": 7716
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 1.5729,
      "step": 7717
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0002,
      "loss": 0.883,
      "step": 7718
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.0704,
      "step": 7719
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 0.9183,
      "step": 7720
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 0.999,
      "step": 7721
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 0.7519,
      "step": 7722
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.0319,
      "step": 7723
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.3797,
      "step": 7724
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.7515,
      "step": 7725
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 0.9664,
      "step": 7726
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.09375,
      "learning_rate": 0.0002,
      "loss": 0.7018,
      "step": 7727
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.1247,
      "step": 7728
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.9949,
      "step": 7729
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.0577,
      "step": 7730
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.1974,
      "step": 7731
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.5172,
      "step": 7732
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.342,
      "step": 7733
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.806,
      "step": 7734
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.2689,
      "step": 7735
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.1449,
      "step": 7736
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.1178,
      "step": 7737
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 0.7354,
      "step": 7738
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.1583,
      "step": 7739
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 0.743,
      "step": 7740
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.124,
      "step": 7741
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.0574,
      "step": 7742
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.9256,
      "step": 7743
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.0247,
      "step": 7744
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.0464,
      "step": 7745
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.0307,
      "step": 7746
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.2503,
      "step": 7747
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0002,
      "loss": 1.2947,
      "step": 7748
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 0.8527,
      "step": 7749
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.1782,
      "step": 7750
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.7079,
      "step": 7751
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.4023,
      "step": 7752
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.2673,
      "step": 7753
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 0.5971,
      "step": 7754
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 0.9119,
      "step": 7755
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.6578,
      "step": 7756
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.3014,
      "step": 7757
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.0497,
      "step": 7758
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.0018,
      "step": 7759
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.7816,
      "step": 7760
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.2856,
      "step": 7761
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.2616,
      "step": 7762
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.0007,
      "step": 7763
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.2674,
      "step": 7764
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.6837,
      "step": 7765
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.4227,
      "step": 7766
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.445,
      "step": 7767
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.8975,
      "step": 7768
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.0313,
      "step": 7769
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.0885,
      "step": 7770
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.5611,
      "step": 7771
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 0.874,
      "step": 7772
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.0537,
      "step": 7773
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.0617,
      "step": 7774
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 1.2264,
      "step": 7775
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.1188,
      "step": 7776
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.3006,
      "step": 7777
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 1.6212,
      "step": 7778
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 0.3213,
      "step": 7779
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.1412,
      "step": 7780
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 0.973,
      "step": 7781
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.635,
      "step": 7782
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.3244,
      "step": 7783
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.1561,
      "step": 7784
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.3813,
      "step": 7785
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.1056,
      "step": 7786
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.2275,
      "step": 7787
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.0625,
      "step": 7788
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 0.8831,
      "step": 7789
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.1598,
      "step": 7790
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.2006,
      "step": 7791
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.2095,
      "step": 7792
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.773,
      "step": 7793
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.1441,
      "step": 7794
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.9694,
      "step": 7795
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.9876,
      "step": 7796
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.0768,
      "step": 7797
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.1096,
      "step": 7798
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 1.1098,
      "step": 7799
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.6946,
      "step": 7800
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.2749,
      "step": 7801
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.0981,
      "step": 7802
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.4023,
      "step": 7803
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002,
      "loss": 1.1065,
      "step": 7804
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.1416,
      "step": 7805
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.0327,
      "step": 7806
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.2181,
      "step": 7807
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.2567,
      "step": 7808
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.1983,
      "step": 7809
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.049,
      "step": 7810
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.3236,
      "step": 7811
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 0.9598,
      "step": 7812
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.0202,
      "step": 7813
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0002,
      "loss": 1.0612,
      "step": 7814
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.0772,
      "step": 7815
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 0.9701,
      "step": 7816
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.0559,
      "step": 7817
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.1376,
      "step": 7818
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.0505,
      "step": 7819
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.9433,
      "step": 7820
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 0.7043,
      "step": 7821
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.2872,
      "step": 7822
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 1.1531,
      "step": 7823
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 1.1027,
      "step": 7824
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 0.8529,
      "step": 7825
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.1381,
      "step": 7826
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 0.9938,
      "step": 7827
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.1159,
      "step": 7828
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.3366,
      "step": 7829
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.1415,
      "step": 7830
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 1.3357,
      "step": 7831
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.667,
      "step": 7832
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.4688,
      "step": 7833
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0002,
      "loss": 0.8957,
      "step": 7834
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.9829,
      "step": 7835
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.4426,
      "step": 7836
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.1477,
      "step": 7837
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0002,
      "loss": 0.8591,
      "step": 7838
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.9873,
      "step": 7839
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.1728,
      "step": 7840
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.8714,
      "step": 7841
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.1831,
      "step": 7842
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.2301,
      "step": 7843
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 1.3145,
      "step": 7844
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 0.9399,
      "step": 7845
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.6882,
      "step": 7846
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.3341,
      "step": 7847
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.3817,
      "step": 7848
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 1.0548,
      "step": 7849
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 1.3829,
      "step": 7850
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3006,
      "step": 7851
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0438,
      "step": 7852
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.0039,
      "step": 7853
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.0403,
      "step": 7854
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 0.9052,
      "step": 7855
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.2208,
      "step": 7856
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.0449,
      "step": 7857
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.3748,
      "step": 7858
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.9268,
      "step": 7859
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.1254,
      "step": 7860
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 1.2301,
      "step": 7861
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 0.823,
      "step": 7862
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.4041,
      "step": 7863
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.1027,
      "step": 7864
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.1966,
      "step": 7865
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.8418,
      "step": 7866
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 0.7519,
      "step": 7867
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.1632,
      "step": 7868
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.029,
      "step": 7869
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.079,
      "step": 7870
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.9865,
      "step": 7871
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.1412,
      "step": 7872
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0002,
      "loss": 0.8358,
      "step": 7873
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.0434,
      "step": 7874
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 0.9246,
      "step": 7875
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0002,
      "loss": 0.9284,
      "step": 7876
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 0.9055,
      "step": 7877
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 0.645,
      "step": 7878
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.0943,
      "step": 7879
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.7032,
      "step": 7880
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 0.8907,
      "step": 7881
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 1.0566,
      "step": 7882
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.6975,
      "step": 7883
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 1.0669,
      "step": 7884
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.2377,
      "step": 7885
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.2501,
      "step": 7886
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 1.2708,
      "step": 7887
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.124,
      "step": 7888
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.9566,
      "step": 7889
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 0.3376,
      "step": 7890
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.6524,
      "step": 7891
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.9476,
      "step": 7892
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 0.9913,
      "step": 7893
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0002,
      "loss": 0.8029,
      "step": 7894
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 0.8019,
      "step": 7895
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 0.9403,
      "step": 7896
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0002,
      "loss": 0.7543,
      "step": 7897
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.7947,
      "step": 7898
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.0097,
      "step": 7899
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 0.9956,
      "step": 7900
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.3878,
      "step": 7901
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.1381,
      "step": 7902
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 1.0972,
      "step": 7903
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0002,
      "loss": 0.9106,
      "step": 7904
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.3962,
      "step": 7905
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.3761,
      "step": 7906
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 0.9006,
      "step": 7907
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 0.8291,
      "step": 7908
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 0.8083,
      "step": 7909
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.7354,
      "step": 7910
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.1996,
      "step": 7911
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.2552,
      "step": 7912
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.1626,
      "step": 7913
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 0.8701,
      "step": 7914
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.0735,
      "step": 7915
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002,
      "loss": 0.7235,
      "step": 7916
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 0.8882,
      "step": 7917
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.2608,
      "step": 7918
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0002,
      "loss": 1.1153,
      "step": 7919
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.0219,
      "step": 7920
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.9699,
      "step": 7921
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.4151,
      "step": 7922
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.1212,
      "step": 7923
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.2016,
      "step": 7924
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.244,
      "step": 7925
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.4886,
      "step": 7926
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 0.9554,
      "step": 7927
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 0.7323,
      "step": 7928
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.9822,
      "step": 7929
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.1608,
      "step": 7930
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.0555,
      "step": 7931
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.9189,
      "step": 7932
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.1473,
      "step": 7933
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.9377,
      "step": 7934
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.078125,
      "learning_rate": 0.0002,
      "loss": 0.4633,
      "step": 7935
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 1.4668,
      "step": 7936
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.0871,
      "step": 7937
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 1.2981,
      "step": 7938
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 0.8148,
      "step": 7939
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.3197,
      "step": 7940
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.0975,
      "step": 7941
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.982,
      "step": 7942
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.3268,
      "step": 7943
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.2395,
      "step": 7944
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.4082,
      "step": 7945
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.029,
      "step": 7946
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.1902,
      "step": 7947
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.8551,
      "step": 7948
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.1002,
      "step": 7949
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.2246,
      "step": 7950
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.8622,
      "step": 7951
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 0.7872,
      "step": 7952
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.7053,
      "step": 7953
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 0.8971,
      "step": 7954
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.067,
      "step": 7955
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 0.8714,
      "step": 7956
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 1.2702,
      "step": 7957
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.1725,
      "step": 7958
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002,
      "loss": 0.7878,
      "step": 7959
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 1.5206,
      "step": 7960
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.841,
      "step": 7961
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 0.6981,
      "step": 7962
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.0533,
      "step": 7963
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.2712,
      "step": 7964
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 0.9936,
      "step": 7965
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 1.1653,
      "step": 7966
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.08,
      "step": 7967
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 0.9716,
      "step": 7968
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.7732,
      "step": 7969
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.2995,
      "step": 7970
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.203,
      "step": 7971
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 1.4441,
      "step": 7972
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.7925,
      "step": 7973
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 0.292,
      "step": 7974
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 0.9841,
      "step": 7975
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.7125,
      "step": 7976
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0002,
      "loss": 1.456,
      "step": 7977
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.9458,
      "step": 7978
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.1708,
      "step": 7979
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.236,
      "step": 7980
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.2568,
      "step": 7981
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 1.3392,
      "step": 7982
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.0616,
      "step": 7983
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.7433,
      "step": 7984
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 1.1277,
      "step": 7985
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.145,
      "step": 7986
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0002,
      "loss": 0.957,
      "step": 7987
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 1.1437,
      "step": 7988
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.3308,
      "step": 7989
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 1.2308,
      "step": 7990
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.3457,
      "step": 7991
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.0386,
      "step": 7992
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.6137,
      "step": 7993
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.2101,
      "step": 7994
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.2001,
      "step": 7995
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 1.1049,
      "step": 7996
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 0.8341,
      "step": 7997
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 0.8667,
      "step": 7998
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.0439,
      "step": 7999
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.116,
      "step": 8000
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.9913,
      "step": 8001
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 0.9706,
      "step": 8002
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 0.6157,
      "step": 8003
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 0.9784,
      "step": 8004
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.9517,
      "step": 8005
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0002,
      "loss": 0.8421,
      "step": 8006
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.2414,
      "step": 8007
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0002,
      "loss": 0.7608,
      "step": 8008
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002,
      "loss": 1.094,
      "step": 8009
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 0.7303,
      "step": 8010
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.3481,
      "step": 8011
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0002,
      "loss": 0.7036,
      "step": 8012
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.0449,
      "step": 8013
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.537,
      "step": 8014
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0002,
      "loss": 1.0843,
      "step": 8015
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.839,
      "step": 8016
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.9024,
      "step": 8017
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0002,
      "loss": 1.1373,
      "step": 8018
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 1.1944,
      "step": 8019
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.85,
      "step": 8020
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 0.9944,
      "step": 8021
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 1.0585,
      "step": 8022
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.9794,
      "step": 8023
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 1.3376,
      "step": 8024
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.234375,
      "learning_rate": 0.0002,
      "loss": 1.2287,
      "step": 8025
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.9608,
      "step": 8026
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.0494,
      "step": 8027
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 1.4098,
      "step": 8028
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0002,
      "loss": 0.9782,
      "step": 8029
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.0044,
      "step": 8030
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 1.0888,
      "step": 8031
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 0.9071,
      "step": 8032
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0002,
      "loss": 1.1252,
      "step": 8033
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.1628,
      "step": 8034
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 0.831,
      "step": 8035
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002,
      "loss": 1.2676,
      "step": 8036
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.9736,
      "step": 8037
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 0.9811,
      "step": 8038
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 1.0355,
      "step": 8039
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.119,
      "step": 8040
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.0312,
      "step": 8041
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.3378,
      "step": 8042
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.1127,
      "step": 8043
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.3722,
      "step": 8044
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.1213,
      "step": 8045
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.1903,
      "step": 8046
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0002,
      "loss": 0.6207,
      "step": 8047
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.9379,
      "step": 8048
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 0.8726,
      "step": 8049
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0002,
      "loss": 0.9021,
      "step": 8050
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.3325,
      "step": 8051
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0002,
      "loss": 1.224,
      "step": 8052
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 0.6787,
      "step": 8053
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 0.9516,
      "step": 8054
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.3319,
      "step": 8055
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.1926,
      "step": 8056
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 0.9538,
      "step": 8057
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002,
      "loss": 1.3988,
      "step": 8058
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 1.2358,
      "step": 8059
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.0439,
      "step": 8060
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 0.7367,
      "step": 8061
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 0.9283,
      "step": 8062
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.347,
      "step": 8063
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 1.2382,
      "step": 8064
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 1.2581,
      "step": 8065
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 0.9105,
      "step": 8066
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.6225,
      "step": 8067
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0002,
      "loss": 1.2845,
      "step": 8068
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 1.2302,
      "step": 8069
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0002,
      "loss": 1.2913,
      "step": 8070
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0002,
      "loss": 1.0046,
      "step": 8071
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0002,
      "loss": 0.8603,
      "step": 8072
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.734375,
      "learning_rate": 0.0002,
      "loss": 0.8341,
      "step": 8073
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 1.2468,
      "step": 8074
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 0.9868,
      "step": 8075
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 1.0808,
      "step": 8076
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 0.9812,
      "step": 8077
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0002,
      "loss": 1.0649,
      "step": 8078
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0002,
      "loss": 1.3865,
      "step": 8079
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.140625,
      "learning_rate": 0.0002,
      "loss": 0.6042,
      "step": 8080
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 1.0939,
      "step": 8081
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 1.2286,
      "step": 8082
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0002,
      "loss": 0.9161,
      "step": 8083
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 1.037,
      "step": 8084
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0002,
      "loss": 1.1794,
      "step": 8085
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 1.2221,
      "step": 8086
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.4593,
      "step": 8087
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.1726,
      "step": 8088
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0002,
      "loss": 1.2258,
      "step": 8089
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 1.089,
      "step": 8090
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.091,
      "step": 8091
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 1.0167,
      "step": 8092
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.8554,
      "step": 8093
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 1.0539,
      "step": 8094
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002,
      "loss": 1.3996,
      "step": 8095
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0002,
      "loss": 0.7073,
      "step": 8096
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 1.4831,
      "step": 8097
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0002,
      "loss": 0.7745,
      "step": 8098
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0002,
      "loss": 1.3726,
      "step": 8099
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0002,
      "loss": 1.1741,
      "step": 8100
    }
  ],
  "logging_steps": 1,
  "max_steps": 18225,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 100,
  "total_flos": 8.26233971539968e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}