diff --git "a/checkpoint-919/trainer_state.json" "b/checkpoint-919/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-919/trainer_state.json"
@@ -0,0 +1,5533 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 919,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.777777777777778e-06,
+      "loss": 3.1281,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.555555555555556e-06,
+      "loss": 3.2134,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 3.1329,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1111111111111112e-05,
+      "loss": 3.1304,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3888888888888888e-05,
+      "loss": 3.1683,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 3.2336,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9444444444444445e-05,
+      "loss": 3.2424,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2222222222222223e-05,
+      "loss": 3.2155,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-05,
+      "loss": 3.1961,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7777777777777776e-05,
+      "loss": 3.2881,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0555555555555554e-05,
+      "loss": 3.1787,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 3.2492,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.611111111111111e-05,
+      "loss": 3.0484,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.888888888888889e-05,
+      "loss": 3.0527,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1666666666666665e-05,
+      "loss": 3.1198,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 3.0216,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.722222222222222e-05,
+      "loss": 2.9886,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-05,
+      "loss": 2.9654,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.277777777777778e-05,
+      "loss": 2.9086,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.555555555555555e-05,
+      "loss": 2.948,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.833333333333333e-05,
+      "loss": 2.9377,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.111111111111111e-05,
+      "loss": 2.9113,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.388888888888888e-05,
+      "loss": 2.7984,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 2.7287,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.944444444444444e-05,
+      "loss": 2.6983,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.222222222222222e-05,
+      "loss": 2.6575,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.5e-05,
+      "loss": 2.6286,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.777777777777778e-05,
+      "loss": 2.5648,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.055555555555556e-05,
+      "loss": 2.5435,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.333333333333333e-05,
+      "loss": 2.5417,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.611111111111112e-05,
+      "loss": 2.4647,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.888888888888889e-05,
+      "loss": 2.4418,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.166666666666667e-05,
+      "loss": 2.3412,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.444444444444444e-05,
+      "loss": 2.3532,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.722222222222223e-05,
+      "loss": 2.2995,
+      "step": 35
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001,
+      "loss": 2.3202,
+      "step": 36
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010277777777777778,
+      "loss": 2.1677,
+      "step": 37
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010555555555555555,
+      "loss": 2.1255,
+      "step": 38
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00010833333333333334,
+      "loss": 2.0289,
+      "step": 39
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001111111111111111,
+      "loss": 2.0142,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011388888888888889,
+      "loss": 1.998,
+      "step": 41
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011666666666666667,
+      "loss": 1.9188,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011944444444444445,
+      "loss": 1.8715,
+      "step": 43
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012222222222222221,
+      "loss": 1.8904,
+      "step": 44
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000125,
+      "loss": 1.8716,
+      "step": 45
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012777777777777776,
+      "loss": 1.7882,
+      "step": 46
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013055555555555558,
+      "loss": 1.7679,
+      "step": 47
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 1.7151,
+      "step": 48
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001361111111111111,
+      "loss": 1.6912,
+      "step": 49
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001388888888888889,
+      "loss": 1.7754,
+      "step": 50
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014166666666666668,
+      "loss": 1.683,
+      "step": 51
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014444444444444444,
+      "loss": 1.6863,
+      "step": 52
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014722222222222223,
+      "loss": 1.6517,
+      "step": 53
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015,
+      "loss": 1.6864,
+      "step": 54
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001527777777777778,
+      "loss": 1.6162,
+      "step": 55
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015555555555555556,
+      "loss": 1.5718,
+      "step": 56
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00015833333333333332,
+      "loss": 1.5966,
+      "step": 57
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001611111111111111,
+      "loss": 1.5594,
+      "step": 58
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001638888888888889,
+      "loss": 1.592,
+      "step": 59
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016666666666666666,
+      "loss": 1.5515,
+      "step": 60
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016944444444444445,
+      "loss": 1.606,
+      "step": 61
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017222222222222224,
+      "loss": 1.5634,
+      "step": 62
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000175,
+      "loss": 1.5984,
+      "step": 63
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00017777777777777779,
+      "loss": 1.577,
+      "step": 64
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018055555555555555,
+      "loss": 1.58,
+      "step": 65
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018333333333333334,
+      "loss": 1.4992,
+      "step": 66
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018611111111111112,
+      "loss": 1.5873,
+      "step": 67
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00018888888888888888,
+      "loss": 1.5352,
+      "step": 68
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019166666666666667,
+      "loss": 1.5,
+      "step": 69
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019444444444444446,
+      "loss": 1.5277,
+      "step": 70
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019722222222222222,
+      "loss": 1.5331,
+      "step": 71
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 1.5179,
+      "step": 72
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020277777777777777,
+      "loss": 1.4606,
+      "step": 73
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020555555555555556,
+      "loss": 1.4401,
+      "step": 74
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00020833333333333335,
+      "loss": 1.5281,
+      "step": 75
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002111111111111111,
+      "loss": 1.4341,
+      "step": 76
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002138888888888889,
+      "loss": 1.4533,
+      "step": 77
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00021666666666666668,
+      "loss": 1.4776,
+      "step": 78
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00021944444444444444,
+      "loss": 1.5066,
+      "step": 79
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002222222222222222,
+      "loss": 1.5293,
+      "step": 80
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022500000000000002,
+      "loss": 1.4638,
+      "step": 81
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00022777777777777778,
+      "loss": 1.4485,
+      "step": 82
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00023055555555555557,
+      "loss": 1.5267,
+      "step": 83
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00023333333333333333,
+      "loss": 1.5055,
+      "step": 84
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00023611111111111112,
+      "loss": 1.4261,
+      "step": 85
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002388888888888889,
+      "loss": 1.3983,
+      "step": 86
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00024166666666666667,
+      "loss": 1.4061,
+      "step": 87
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024444444444444443,
+      "loss": 1.4345,
+      "step": 88
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00024722222222222224,
+      "loss": 1.4546,
+      "step": 89
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025,
+      "loss": 1.3898,
+      "step": 90
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025277777777777777,
+      "loss": 1.4346,
+      "step": 91
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025555555555555553,
+      "loss": 1.4608,
+      "step": 92
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00025833333333333334,
+      "loss": 1.4902,
+      "step": 93
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00026111111111111116,
+      "loss": 1.398,
+      "step": 94
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002638888888888889,
+      "loss": 1.4108,
+      "step": 95
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002666666666666667,
+      "loss": 1.3904,
+      "step": 96
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00026944444444444444,
+      "loss": 1.3856,
+      "step": 97
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002722222222222222,
+      "loss": 1.412,
+      "step": 98
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000275,
+      "loss": 1.4201,
+      "step": 99
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002777777777777778,
+      "loss": 1.3526,
+      "step": 100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00028055555555555554,
+      "loss": 1.4473,
+      "step": 101
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00028333333333333335,
+      "loss": 1.3979,
+      "step": 102
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002861111111111111,
+      "loss": 1.3605,
+      "step": 103
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002888888888888889,
+      "loss": 1.389,
+      "step": 104
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002916666666666667,
+      "loss": 1.3683,
+      "step": 105
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029444444444444445,
+      "loss": 1.3876,
+      "step": 106
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002972222222222222,
+      "loss": 1.4033,
+      "step": 107
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003,
+      "loss": 1.3951,
+      "step": 108
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003027777777777778,
+      "loss": 1.3494,
+      "step": 109
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003055555555555556,
+      "loss": 1.3452,
+      "step": 110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00030833333333333337,
+      "loss": 1.4116,
+      "step": 111
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003111111111111111,
+      "loss": 1.4196,
+      "step": 112
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003138888888888889,
+      "loss": 1.3441,
+      "step": 113
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00031666666666666665,
+      "loss": 1.3614,
+      "step": 114
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003194444444444444,
+      "loss": 1.3536,
+      "step": 115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003222222222222222,
+      "loss": 1.3626,
+      "step": 116
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00032500000000000004,
+      "loss": 1.3436,
+      "step": 117
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003277777777777778,
+      "loss": 1.3135,
+      "step": 118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00033055555555555556,
+      "loss": 1.3916,
+      "step": 119
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003333333333333333,
+      "loss": 1.3465,
+      "step": 120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00033611111111111114,
+      "loss": 1.3628,
+      "step": 121
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003388888888888889,
+      "loss": 1.4076,
+      "step": 122
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00034166666666666666,
+      "loss": 1.2934,
+      "step": 123
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003444444444444445,
+      "loss": 1.3353,
+      "step": 124
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00034722222222222224,
+      "loss": 1.2837,
+      "step": 125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035,
+      "loss": 1.2845,
+      "step": 126
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003527777777777778,
+      "loss": 1.2902,
+      "step": 127
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035555555555555557,
+      "loss": 1.3844,
+      "step": 128
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00035833333333333333,
+      "loss": 1.3711,
+      "step": 129
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003611111111111111,
+      "loss": 1.3353,
+      "step": 130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00036388888888888886,
+      "loss": 1.3262,
+      "step": 131
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00036666666666666667,
+      "loss": 1.2868,
+      "step": 132
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003694444444444445,
+      "loss": 1.3332,
+      "step": 133
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00037222222222222225,
+      "loss": 1.2785,
+      "step": 134
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000375,
+      "loss": 1.3316,
+      "step": 135
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00037777777777777777,
+      "loss": 1.3902,
+      "step": 136
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038055555555555553,
+      "loss": 1.3159,
+      "step": 137
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038333333333333334,
+      "loss": 1.3355,
+      "step": 138
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038611111111111116,
+      "loss": 1.2676,
+      "step": 139
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003888888888888889,
+      "loss": 1.2722,
+      "step": 140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003916666666666667,
+      "loss": 1.3487,
+      "step": 141
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039444444444444444,
+      "loss": 1.3746,
+      "step": 142
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003972222222222222,
+      "loss": 1.2832,
+      "step": 143
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004,
+      "loss": 1.3174,
+      "step": 144
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004027777777777778,
+      "loss": 1.3028,
+      "step": 145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040555555555555554,
+      "loss": 1.2666,
+      "step": 146
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00040833333333333336,
+      "loss": 1.2877,
+      "step": 147
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004111111111111111,
+      "loss": 1.2522,
+      "step": 148
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004138888888888889,
+      "loss": 1.3421,
+      "step": 149
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004166666666666667,
+      "loss": 1.2411,
+      "step": 150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00041944444444444445,
+      "loss": 1.2629,
+      "step": 151
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004222222222222222,
+      "loss": 1.2925,
+      "step": 152
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000425,
+      "loss": 1.2424,
+      "step": 153
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004277777777777778,
+      "loss": 1.2099,
+      "step": 154
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004305555555555556,
+      "loss": 1.232,
+      "step": 155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00043333333333333337,
+      "loss": 1.3073,
+      "step": 156
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00043611111111111113,
+      "loss": 1.2503,
+      "step": 157
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004388888888888889,
+      "loss": 1.2967,
+      "step": 158
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00044166666666666665,
+      "loss": 1.2743,
+      "step": 159
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004444444444444444,
+      "loss": 1.2235,
+      "step": 160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004472222222222222,
+      "loss": 1.2514,
+      "step": 161
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045000000000000004,
+      "loss": 1.2758,
+      "step": 162
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004527777777777778,
+      "loss": 1.2982,
+      "step": 163
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00045555555555555556,
+      "loss": 1.3149,
+      "step": 164
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004583333333333333,
+      "loss": 1.2579,
+      "step": 165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046111111111111114,
+      "loss": 1.2657,
+      "step": 166
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004638888888888889,
+      "loss": 1.2372,
+      "step": 167
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046666666666666666,
+      "loss": 1.203,
+      "step": 168
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004694444444444445,
+      "loss": 1.2448,
+      "step": 169
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00047222222222222224,
+      "loss": 1.25,
+      "step": 170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000475,
+      "loss": 1.2973,
+      "step": 171
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004777777777777778,
+      "loss": 1.2044,
+      "step": 172
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004805555555555556,
+      "loss": 1.1807,
+      "step": 173
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00048333333333333334,
+      "loss": 1.2447,
+      "step": 174
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004861111111111111,
+      "loss": 1.265,
+      "step": 175
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004888888888888889,
+      "loss": 1.2603,
+      "step": 176
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004916666666666666,
+      "loss": 1.2455,
+      "step": 177
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004944444444444445,
+      "loss": 1.1926,
+      "step": 178
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004972222222222222,
+      "loss": 1.2319,
+      "step": 179
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005,
+      "loss": 1.2501,
+      "step": 180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005,
+      "loss": 1.2104,
+      "step": 181
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005,
+      "loss": 1.1955,
+      "step": 182
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005,
+      "loss": 1.1759,
+      "step": 183
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005,
+      "loss": 1.2627,
+      "step": 184
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005,
+      "loss": 1.1659,
+      "step": 185
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005,
+      "loss": 1.2082,
+      "step": 186
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005,
+      "loss": 1.2842,
+      "step": 187
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0005,
+      "loss": 1.1943,
+      "step": 188
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005,
+      "loss": 1.2494,
+      "step": 189
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005,
+      "loss": 1.1935,
+      "step": 190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005,
+      "loss": 1.1754,
+      "step": 191
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005,
+      "loss": 1.2717,
+      "step": 192
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005,
+      "loss": 1.2008,
+      "step": 193
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005,
+      "loss": 1.2049,
+      "step": 194
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005,
+      "loss": 1.182,
+      "step": 195
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005,
+      "loss": 1.1936,
+      "step": 196
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0005,
+      "loss": 1.2024,
+      "step": 197
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005,
+      "loss": 1.2534,
+      "step": 198
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005,
+      "loss": 1.1986,
+      "step": 199
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005,
+      "loss": 1.3031,
+      "step": 200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005,
+      "loss": 1.2503,
+      "step": 201
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005,
+      "loss": 1.1916,
+      "step": 202
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005,
+      "loss": 1.1704,
+      "step": 203
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005,
+      "loss": 1.2644,
+      "step": 204
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005,
+      "loss": 1.1649,
+      "step": 205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0005,
+      "loss": 1.1897,
+      "step": 206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005,
+      "loss": 1.2378,
+      "step": 207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005,
+      "loss": 1.1909,
+      "step": 208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005,
+      "loss": 1.1687,
+      "step": 209
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005,
+      "loss": 1.1617,
+      "step": 210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005,
+      "loss": 1.1845,
+      "step": 211
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005,
+      "loss": 1.2246,
+      "step": 212
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005,
+      "loss": 1.1967,
+      "step": 213
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005,
+      "loss": 1.1896,
+      "step": 214
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005,
+      "loss": 1.2793,
+      "step": 215
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005,
+      "loss": 1.2362,
+      "step": 216
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005,
+      "loss": 1.1963,
+      "step": 217
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005,
+      "loss": 1.1692,
+      "step": 218
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005,
+      "loss": 1.215,
+      "step": 219
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005,
+      "loss": 1.1741,
+      "step": 220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005,
+      "loss": 1.2078,
+      "step": 221
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005,
+      "loss": 1.1777,
+      "step": 222
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005,
+      "loss": 1.2123,
+      "step": 223
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005,
+      "loss": 1.1975,
+      "step": 224
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005,
+      "loss": 1.2305,
+      "step": 225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005,
+      "loss": 1.1652,
+      "step": 226
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005,
+      "loss": 1.1781,
+      "step": 227
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005,
+      "loss": 1.1968,
+      "step": 228
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005,
+      "loss": 1.1425,
+      "step": 229
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005,
+      "loss": 1.1933,
+      "step": 230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005,
+      "loss": 1.151,
+      "step": 231
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005,
+      "loss": 1.1923,
+      "step": 232
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005,
+      "loss": 1.2289,
+      "step": 233
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005,
+      "loss": 1.1526,
+      "step": 234
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005,
+      "loss": 1.2487,
+      "step": 235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005,
+      "loss": 1.1658,
+      "step": 236
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005,
+      "loss": 1.1752,
+      "step": 237
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005,
+      "loss": 1.2331,
+      "step": 238
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005,
+      "loss": 1.2081,
+      "step": 239
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005,
+      "loss": 1.1688,
+      "step": 240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005,
+      "loss": 1.1753,
+      "step": 241
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005,
+      "loss": 1.1844,
+      "step": 242
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005,
+      "loss": 1.233,
+      "step": 243
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005,
+      "loss": 1.1607,
+      "step": 244
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005,
+      "loss": 1.1212,
+      "step": 245
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005,
+      "loss": 1.1657,
+      "step": 246
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005,
+      "loss": 1.1642,
+      "step": 247
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005,
+      "loss": 1.2154,
+      "step": 248
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005,
+      "loss": 1.1691,
+      "step": 249
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005,
+      "loss": 1.1813,
+      "step": 250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005,
+      "loss": 1.1523,
+      "step": 251
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0005,
+      "loss": 1.1322,
+      "step": 252
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005,
+      "loss": 1.2245,
+      "step": 253
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005,
+      "loss": 1.1752,
+      "step": 254
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005,
+      "loss": 1.1596,
+      "step": 255
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005,
+      "loss": 1.1393,
+      "step": 256
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005,
+      "loss": 1.1568,
+      "step": 257
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005,
+      "loss": 1.188,
+      "step": 258
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005,
+      "loss": 1.152,
+      "step": 259
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005,
+      "loss": 1.1747,
+      "step": 260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0005,
+      "loss": 1.1492,
+      "step": 261
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005,
+      "loss": 1.1867,
+      "step": 262
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005,
+      "loss": 1.1415,
+      "step": 263
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005,
+      "loss": 1.0918,
+      "step": 264
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005,
+      "loss": 1.1829,
+      "step": 265
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005,
+      "loss": 1.1251,
+      "step": 266
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005,
+      "loss": 1.1948,
+      "step": 267
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005,
+      "loss": 1.0871,
+      "step": 268
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005,
+      "loss": 1.1583,
+      "step": 269
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005,
+      "loss": 1.1035,
+      "step": 270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0005,
+      "loss": 1.2108,
+      "step": 271
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005,
+      "loss": 1.1617,
+      "step": 272
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005,
+      "loss": 1.1415,
+      "step": 273
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005,
+      "loss": 1.1835,
+      "step": 274
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005,
+      "loss": 1.1669,
+      "step": 275
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005,
+      "loss": 1.1589,
+      "step": 276
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005,
+      "loss": 1.1551,
+      "step": 277
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005,
+      "loss": 1.1896,
+      "step": 278
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005,
+      "loss": 1.1635,
+      "step": 279
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0005,
+      "loss": 1.1318,
+      "step": 280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005,
+      "loss": 1.1901,
+      "step": 281
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005,
+      "loss": 1.1671,
+      "step": 282
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005,
+      "loss": 1.1711,
+      "step": 283
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005,
+      "loss": 1.121,
+      "step": 284
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005,
+      "loss": 1.1899,
+      "step": 285
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005,
+      "loss": 1.1769,
+      "step": 286
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005,
+      "loss": 1.1859,
+      "step": 287
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005,
+      "loss": 1.1547,
+      "step": 288
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0005,
+      "loss": 1.0976,
+      "step": 289
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005,
+      "loss": 1.1912,
+      "step": 290
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005,
+      "loss": 1.2091,
+      "step": 291
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005,
+      "loss": 1.1092,
+      "step": 292
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005,
+      "loss": 1.1147,
+      "step": 293
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005,
+      "loss": 1.1473,
+      "step": 294
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005,
+      "loss": 1.1025,
+      "step": 295
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005,
+      "loss": 1.1653,
+      "step": 296
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005,
+      "loss": 1.1267,
+      "step": 297
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0005,
+      "loss": 1.1704,
+      "step": 298
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005,
+      "loss": 1.1654,
+      "step": 299
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005,
+      "loss": 1.17,
+      "step": 300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005,
+      "loss": 1.1459,
+      "step": 301
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005,
+      "loss": 1.154,
+      "step": 302
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005,
+      "loss": 1.1272,
+      "step": 303
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005,
+      "loss": 1.1724,
+      "step": 304
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005,
+      "loss": 1.2047,
+      "step": 305
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005,
+      "loss": 1.1838,
+      "step": 306
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0005,
+      "loss": 1.1326,
+      "step": 307
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005,
+      "loss": 1.1063,
+      "step": 308
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005,
+      "loss": 1.1382,
+      "step": 309
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005,
+      "loss": 1.2094,
+      "step": 310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005,
+      "loss": 1.1924,
+      "step": 311
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005,
+      "loss": 1.1513,
+      "step": 312
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005,
+      "loss": 1.1471,
+      "step": 313
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005,
+      "loss": 1.1268,
+      "step": 314
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005,
+      "loss": 1.1566,
+      "step": 315
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005,
+      "loss": 1.0798,
+      "step": 316
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0005,
+      "loss": 1.267,
+      "step": 317
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005,
+      "loss": 1.1165,
+      "step": 318
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005,
+      "loss": 1.0973,
+      "step": 319
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005,
+      "loss": 1.1077,
+      "step": 320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005,
+      "loss": 1.1538,
+      "step": 321
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005,
+      "loss": 1.1111,
+      "step": 322
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005,
+      "loss": 1.1497,
+      "step": 323
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005,
+      "loss": 1.1579,
+      "step": 324
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005,
+      "loss": 1.1407,
+      "step": 325
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0005,
+      "loss": 1.1359,
+      "step": 326
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005,
+      "loss": 1.1745,
+      "step": 327
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005,
+      "loss": 1.167,
+      "step": 328
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005,
+      "loss": 1.1816,
+      "step": 329
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005,
+      "loss": 1.169,
+      "step": 330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005,
+      "loss": 1.119,
+      "step": 331
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005,
+      "loss": 1.1806,
+      "step": 332
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005,
+      "loss": 1.1916,
+      "step": 333
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005,
+      "loss": 1.1489,
+      "step": 334
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0005,
+      "loss": 1.133,
+      "step": 335
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005,
+      "loss": 1.1741,
+      "step": 336
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005,
+      "loss": 1.1132,
+      "step": 337
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005,
+      "loss": 1.1262,
+      "step": 338
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005,
+      "loss": 1.1072,
+      "step": 339
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005,
+      "loss": 1.124,
+      "step": 340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005,
+      "loss": 1.1135,
+      "step": 341
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005,
+      "loss": 1.1375,
+      "step": 342
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005,
+      "loss": 1.1816,
+      "step": 343
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0005,
+      "loss": 1.1213,
+      "step": 344
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005,
+      "loss": 1.1505,
+      "step": 345
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005,
+      "loss": 1.2095,
+      "step": 346
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005,
+      "loss": 1.1606,
+      "step": 347
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005,
+      "loss": 1.1337,
+      "step": 348
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005,
+      "loss": 1.1569,
+      "step": 349
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005,
+      "loss": 1.159,
+      "step": 350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005,
+      "loss": 1.109,
+      "step": 351
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005,
+      "loss": 1.1253,
+      "step": 352
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0005,
+      "loss": 1.1293,
+      "step": 353
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0005,
+      "loss": 1.1065,
+      "step": 354
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0005,
+      "loss": 1.1155,
+      "step": 355
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0005,
+      "loss": 1.1258,
+      "step": 356
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0005,
+      "loss": 1.1052,
+      "step": 357
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0005,
+      "loss": 1.1212,
+      "step": 358
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0005,
+      "loss": 1.095,
+      "step": 359
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0005,
+      "loss": 1.1834,
+      "step": 360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0005,
+      "loss": 1.1111,
+      "step": 361
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0005,
+      "loss": 1.1592,
+      "step": 362
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0005,
+      "loss": 1.1187,
+      "step": 363
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0005,
+      "loss": 1.1215,
+      "step": 364
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0005,
+      "loss": 1.1184,
+      "step": 365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0005,
+      "loss": 1.0882,
+      "step": 366
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0005,
+      "loss": 1.1468,
+      "step": 367
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0005,
+      "loss": 1.149,
+      "step": 368
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0005,
+      "loss": 1.1485,
+      "step": 369
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0005,
+      "loss": 1.1312,
+      "step": 370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0005,
+      "loss": 1.0817,
+      "step": 371
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0005,
+      "loss": 1.1428,
+      "step": 372
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0005,
+      "loss": 1.0791,
+      "step": 373
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0005,
+      "loss": 1.1449,
+      "step": 374
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0005,
+      "loss": 1.1281,
+      "step": 375
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0005,
+      "loss": 1.1784,
+      "step": 376
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0005,
+      "loss": 1.1194,
+      "step": 377
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0005,
+      "loss": 1.1417,
+      "step": 378
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0005,
+      "loss": 1.1331,
+      "step": 379
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0005,
+      "loss": 1.1249,
+      "step": 380
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0005,
+      "loss": 1.1208,
+      "step": 381
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0005,
+      "loss": 1.1591,
+      "step": 382
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0005,
+      "loss": 1.1562,
+      "step": 383
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0005,
+      "loss": 1.1948,
+      "step": 384
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0005,
+      "loss": 1.1647,
+      "step": 385
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0005,
+      "loss": 1.0564,
+      "step": 386
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0005,
+      "loss": 1.1051,
+      "step": 387
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0005,
+      "loss": 1.11,
+      "step": 388
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0005,
+      "loss": 1.0871,
+      "step": 389
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0005,
+      "loss": 1.0904,
+      "step": 390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0005,
+      "loss": 1.1267,
+      "step": 391
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0005,
+      "loss": 1.1594,
+      "step": 392
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0005,
+      "loss": 1.1525,
+      "step": 393
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0005,
+      "loss": 1.0888,
+      "step": 394
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0005,
+      "loss": 1.1201,
+      "step": 395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0005,
+      "loss": 1.1309,
+      "step": 396
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0005,
+      "loss": 1.1618,
+      "step": 397
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0005,
+      "loss": 1.1327,
+      "step": 398
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0005,
+      "loss": 1.1383,
+      "step": 399
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0005,
+      "loss": 1.1226,
+      "step": 400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0005,
+      "loss": 1.1208,
+      "step": 401
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0005,
+      "loss": 1.1492,
+      "step": 402
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0005,
+      "loss": 1.1441,
+      "step": 403
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0005,
+      "loss": 1.1742,
+      "step": 404
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0005,
+      "loss": 1.1154,
+      "step": 405
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0005,
+      "loss": 1.1704,
+      "step": 406
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0005,
+      "loss": 1.0913,
+      "step": 407
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0005,
+      "loss": 1.1144,
+      "step": 408
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005,
+      "loss": 1.1461,
+      "step": 409
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005,
+      "loss": 1.186,
+      "step": 410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005,
+      "loss": 1.1279,
+      "step": 411
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005,
+      "loss": 1.1027,
+      "step": 412
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005,
+      "loss": 1.0754,
+      "step": 413
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005,
+      "loss": 1.1038,
+      "step": 414
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005,
+      "loss": 1.1766,
+      "step": 415
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005,
+      "loss": 1.1195,
+      "step": 416
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005,
+      "loss": 1.136,
+      "step": 417
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005,
+      "loss": 1.148,
+      "step": 418
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005,
+      "loss": 1.1444,
+      "step": 419
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005,
+      "loss": 1.197,
+      "step": 420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005,
+      "loss": 1.1144,
+      "step": 421
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005,
+      "loss": 1.141,
+      "step": 422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005,
+      "loss": 1.1381,
+      "step": 423
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005,
+      "loss": 1.191,
+      "step": 424
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005,
+      "loss": 1.0779,
+      "step": 425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005,
+      "loss": 1.1022,
+      "step": 426
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005,
+      "loss": 1.0886,
+      "step": 427
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005,
+      "loss": 1.1454,
+      "step": 428
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005,
+      "loss": 1.1257,
+      "step": 429
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005,
+      "loss": 1.0935,
+      "step": 430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005,
+      "loss": 1.0932,
+      "step": 431
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005,
+      "loss": 1.1824,
+      "step": 432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005,
+      "loss": 1.1599,
+      "step": 433
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005,
+      "loss": 1.1213,
+      "step": 434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005,
+      "loss": 1.1425,
+      "step": 435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005,
+      "loss": 1.094,
+      "step": 436
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005,
+      "loss": 1.1431,
+      "step": 437
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005,
+      "loss": 1.1075,
+      "step": 438
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005,
+      "loss": 1.1029,
+      "step": 439
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005,
+      "loss": 1.0991,
+      "step": 440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005,
+      "loss": 1.107,
+      "step": 441
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005,
+      "loss": 1.1108,
+      "step": 442
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005,
+      "loss": 1.1109,
+      "step": 443
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005,
+      "loss": 1.1001,
+      "step": 444
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005,
+      "loss": 1.1122,
+      "step": 445
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005,
+      "loss": 1.1647,
+      "step": 446
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005,
+      "loss": 1.1571,
+      "step": 447
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005,
+      "loss": 1.1203,
+      "step": 448
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005,
+      "loss": 1.1374,
+      "step": 449
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005,
+      "loss": 1.1189,
+      "step": 450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005,
+      "loss": 1.1531,
+      "step": 451
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005,
+      "loss": 1.1362,
+      "step": 452
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005,
+      "loss": 1.1746,
+      "step": 453
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005,
+      "loss": 1.149,
+      "step": 454
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005,
+      "loss": 1.1452,
+      "step": 455
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005,
+      "loss": 1.1029,
+      "step": 456
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005,
+      "loss": 1.1254,
+      "step": 457
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005,
+      "loss": 1.0814,
+      "step": 458
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005,
+      "loss": 1.1769,
+      "step": 459
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005,
+      "loss": 1.1446,
+      "step": 460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005,
+      "loss": 1.1435,
+      "step": 461
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005,
+      "loss": 1.1249,
+      "step": 462
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005,
+      "loss": 1.0767,
+      "step": 463
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005,
+      "loss": 1.0756,
+      "step": 464
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005,
+      "loss": 1.1044,
+      "step": 465
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005,
+      "loss": 1.1442,
+      "step": 466
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005,
+      "loss": 1.1533,
+      "step": 467
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005,
+      "loss": 1.1731,
+      "step": 468
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005,
+      "loss": 1.1051,
+      "step": 469
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005,
+      "loss": 1.1828,
+      "step": 470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005,
+      "loss": 1.1326,
+      "step": 471
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005,
+      "loss": 1.1079,
+      "step": 472
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005,
+      "loss": 1.1553,
+      "step": 473
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0005,
+      "loss": 1.1859,
+      "step": 474
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0005,
+      "loss": 1.1411,
+      "step": 475
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0005,
+      "loss": 1.0923,
+      "step": 476
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0005,
+      "loss": 1.1239,
+      "step": 477
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0005,
+      "loss": 1.18,
+      "step": 478
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0005,
+      "loss": 1.1454,
+      "step": 479
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0005,
+      "loss": 1.1222,
+      "step": 480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0005,
+      "loss": 1.1965,
+      "step": 481
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0005,
+      "loss": 1.1057,
+      "step": 482
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0005,
+      "loss": 1.128,
+      "step": 483
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0005,
+      "loss": 1.13,
+      "step": 484
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0005,
+      "loss": 1.129,
+      "step": 485
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0005,
+      "loss": 1.0847,
+      "step": 486
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0005,
+      "loss": 1.1389,
+      "step": 487
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0005,
+      "loss": 1.1743,
+      "step": 488
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0005,
+      "loss": 1.1498,
+      "step": 489
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0005,
+      "loss": 1.1089,
+      "step": 490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0005,
+      "loss": 1.1288,
+      "step": 491
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0005,
+      "loss": 1.1073,
+      "step": 492
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0005,
+      "loss": 1.1023,
+      "step": 493
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0005,
+      "loss": 1.1778,
+      "step": 494
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0005,
+      "loss": 1.1236,
+      "step": 495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0005,
+      "loss": 1.129,
+      "step": 496
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0005,
+      "loss": 1.1088,
+      "step": 497
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0005,
+      "loss": 1.1641,
+      "step": 498
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0005,
+      "loss": 1.0958,
+      "step": 499
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0005,
+      "loss": 1.0925,
+      "step": 500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0005,
+      "loss": 1.1208,
+      "step": 501
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0005,
+      "loss": 1.1541,
+      "step": 502
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0005,
+      "loss": 1.0887,
+      "step": 503
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0005,
+      "loss": 1.1331,
+      "step": 504
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0005,
+      "loss": 1.1128,
+      "step": 505
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0005,
+      "loss": 1.0945,
+      "step": 506
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0005,
+      "loss": 1.1026,
+      "step": 507
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0005,
+      "loss": 1.1539,
+      "step": 508
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0005,
+      "loss": 1.1286,
+      "step": 509
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0005,
+      "loss": 1.0453,
+      "step": 510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0005,
+      "loss": 1.1376,
+      "step": 511
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0005,
+      "loss": 1.1523,
+      "step": 512
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0005,
+      "loss": 1.0747,
+      "step": 513
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0005,
+      "loss": 1.1524,
+      "step": 514
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0005,
+      "loss": 1.1498,
+      "step": 515
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0005,
+      "loss": 1.1406,
+      "step": 516
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0005,
+      "loss": 1.1667,
+      "step": 517
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0005,
+      "loss": 1.1435,
+      "step": 518
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0005,
+      "loss": 1.1327,
+      "step": 519
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0005,
+      "loss": 1.1433,
+      "step": 520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0005,
+      "loss": 1.1048,
+      "step": 521
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0005,
+      "loss": 1.1297,
+      "step": 522
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0005,
+      "loss": 1.1026,
+      "step": 523
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0005,
+      "loss": 1.1478,
+      "step": 524
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0005,
+      "loss": 1.094,
+      "step": 525
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0005,
+      "loss": 1.1168,
+      "step": 526
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0005,
+      "loss": 1.1368,
+      "step": 527
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0005,
+      "loss": 1.1093,
+      "step": 528
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0005,
+      "loss": 1.1417,
+      "step": 529
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0005,
+      "loss": 1.1014,
+      "step": 530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0005,
+      "loss": 1.1483,
+      "step": 531
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0005,
+      "loss": 1.1375,
+      "step": 532
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0005,
+      "loss": 1.1301,
+      "step": 533
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0005,
+      "loss": 1.1577,
+      "step": 534
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0005,
+      "loss": 1.1538,
+      "step": 535
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0005,
+      "loss": 1.1019,
+      "step": 536
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0005,
+      "loss": 1.0672,
+      "step": 537
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0005,
+      "loss": 1.1381,
+      "step": 538
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0005,
+      "loss": 1.1205,
+      "step": 539
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0005,
+      "loss": 1.1437,
+      "step": 540
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0005,
+      "loss": 1.1452,
+      "step": 541
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0005,
+      "loss": 1.1487,
+      "step": 542
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0005,
+      "loss": 1.1483,
+      "step": 543
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0005,
+      "loss": 1.1613,
+      "step": 544
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0005,
+      "loss": 1.171,
+      "step": 545
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0005,
+      "loss": 1.1422,
+      "step": 546
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0005,
+      "loss": 1.0819,
+      "step": 547
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0005,
+      "loss": 1.151,
+      "step": 548
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0005,
+      "loss": 1.1284,
+      "step": 549
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0005,
+      "loss": 1.0728,
+      "step": 550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0005,
+      "loss": 1.1174,
+      "step": 551
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0005,
+      "loss": 1.0932,
+      "step": 552
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0005,
+      "loss": 1.1282,
+      "step": 553
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0005,
+      "loss": 1.1164,
+      "step": 554
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0005,
+      "loss": 1.1183,
+      "step": 555
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0005,
+      "loss": 1.1239,
+      "step": 556
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0005,
+      "loss": 1.158,
+      "step": 557
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0005,
+      "loss": 1.0952,
+      "step": 558
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0005,
+      "loss": 1.1344,
+      "step": 559
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0005,
+      "loss": 1.0638,
+      "step": 560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0005,
+      "loss": 1.1002,
+      "step": 561
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0005,
+      "loss": 1.1048,
+      "step": 562
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0005,
+      "loss": 1.1261,
+      "step": 563
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0005,
+      "loss": 1.1568,
+      "step": 564
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0005,
+      "loss": 1.1359,
+      "step": 565
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0005,
+      "loss": 1.1488,
+      "step": 566
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0005,
+      "loss": 1.059,
+      "step": 567
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0005,
+      "loss": 1.1194,
+      "step": 568
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0005,
+      "loss": 1.0967,
+      "step": 569
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0005,
+      "loss": 1.1654,
+      "step": 570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0005,
+      "loss": 1.1046,
+      "step": 571
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0005,
+      "loss": 1.1401,
+      "step": 572
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0005,
+      "loss": 1.1222,
+      "step": 573
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0005,
+      "loss": 1.1122,
+      "step": 574
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0005,
+      "loss": 1.088,
+      "step": 575
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0005,
+      "loss": 1.1294,
+      "step": 576
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0005,
+      "loss": 1.0765,
+      "step": 577
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0005,
+      "loss": 1.1138,
+      "step": 578
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0005,
+      "loss": 1.1306,
+      "step": 579
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0005,
+      "loss": 1.1253,
+      "step": 580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0005,
+      "loss": 1.1401,
+      "step": 581
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0005,
+      "loss": 1.1404,
+      "step": 582
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0005,
+      "loss": 1.1131,
+      "step": 583
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005,
+      "loss": 1.0914,
+      "step": 584
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005,
+      "loss": 1.0828,
+      "step": 585
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005,
+      "loss": 1.1405,
+      "step": 586
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005,
+      "loss": 1.1837,
+      "step": 587
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005,
+      "loss": 1.1506,
+      "step": 588
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005,
+      "loss": 1.1738,
+      "step": 589
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005,
+      "loss": 1.1206,
+      "step": 590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005,
+      "loss": 1.0608,
+      "step": 591
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005,
+      "loss": 1.0809,
+      "step": 592
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005,
+      "loss": 1.1042,
+      "step": 593
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005,
+      "loss": 1.1272,
+      "step": 594
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005,
+      "loss": 1.116,
+      "step": 595
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005,
+      "loss": 1.1478,
+      "step": 596
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005,
+      "loss": 1.1259,
+      "step": 597
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005,
+      "loss": 1.1095,
+      "step": 598
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005,
+      "loss": 1.1013,
+      "step": 599
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005,
+      "loss": 1.0996,
+      "step": 600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005,
+      "loss": 1.1599,
+      "step": 601
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005,
+      "loss": 1.0754,
+      "step": 602
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005,
+      "loss": 1.1131,
+      "step": 603
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005,
+      "loss": 1.1147,
+      "step": 604
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005,
+      "loss": 1.0914,
+      "step": 605
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005,
+      "loss": 1.1103,
+      "step": 606
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005,
+      "loss": 1.1215,
+      "step": 607
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005,
+      "loss": 1.1286,
+      "step": 608
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005,
+      "loss": 1.1102,
+      "step": 609
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005,
+      "loss": 1.1272,
+      "step": 610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005,
+      "loss": 1.1174,
+      "step": 611
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005,
+      "loss": 1.1152,
+      "step": 612
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005,
+      "loss": 1.138,
+      "step": 613
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005,
+      "loss": 1.1373,
+      "step": 614
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005,
+      "loss": 1.1462,
+      "step": 615
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005,
+      "loss": 1.1455,
+      "step": 616
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005,
+      "loss": 1.1221,
+      "step": 617
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005,
+      "loss": 1.1226,
+      "step": 618
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005,
+      "loss": 1.1146,
+      "step": 619
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005,
+      "loss": 1.1333,
+      "step": 620
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005,
+      "loss": 1.1346,
+      "step": 621
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005,
+      "loss": 1.1345,
+      "step": 622
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005,
+      "loss": 1.079,
+      "step": 623
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005,
+      "loss": 1.0879,
+      "step": 624
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005,
+      "loss": 1.1246,
+      "step": 625
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005,
+      "loss": 1.1081,
+      "step": 626
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005,
+      "loss": 1.1525,
+      "step": 627
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005,
+      "loss": 1.0974,
+      "step": 628
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005,
+      "loss": 1.1058,
+      "step": 629
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0005,
+      "loss": 1.065,
+      "step": 630
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0005,
+      "loss": 1.1341,
+      "step": 631
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0005,
+      "loss": 1.1271,
+      "step": 632
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0005,
+      "loss": 1.1319,
+      "step": 633
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0005,
+      "loss": 1.0693,
+      "step": 634
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0005,
+      "loss": 1.0998,
+      "step": 635
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0005,
+      "loss": 1.1202,
+      "step": 636
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0005,
+      "loss": 1.1288,
+      "step": 637
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0005,
+      "loss": 1.0911,
+      "step": 638
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0005,
+      "loss": 1.1047,
+      "step": 639
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0005,
+      "loss": 1.1118,
+      "step": 640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0005,
+      "loss": 1.1363,
+      "step": 641
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0005,
+      "loss": 1.1063,
+      "step": 642
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0005,
+      "loss": 1.0746,
+      "step": 643
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0005,
+      "loss": 1.1267,
+      "step": 644
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0005,
+      "loss": 1.1338,
+      "step": 645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0005,
+      "loss": 1.1241,
+      "step": 646
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0005,
+      "loss": 1.0913,
+      "step": 647
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005,
+      "loss": 1.1116,
+      "step": 648
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005,
+      "loss": 1.1093,
+      "step": 649
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005,
+      "loss": 1.0802,
+      "step": 650
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005,
+      "loss": 1.0975,
+      "step": 651
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005,
+      "loss": 1.1244,
+      "step": 652
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005,
+      "loss": 1.107,
+      "step": 653
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005,
+      "loss": 1.074,
+      "step": 654
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005,
+      "loss": 1.1133,
+      "step": 655
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005,
+      "loss": 1.1446,
+      "step": 656
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005,
+      "loss": 1.1164,
+      "step": 657
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0005,
+      "loss": 1.0897,
+      "step": 658
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0005,
+      "loss": 1.1515,
+      "step": 659
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0005,
+      "loss": 1.0802,
+      "step": 660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0005,
+      "loss": 1.126,
+      "step": 661
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0005,
+      "loss": 1.1353,
+      "step": 662
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0005,
+      "loss": 1.0922,
+      "step": 663
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0005,
+      "loss": 1.0776,
+      "step": 664
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0005,
+      "loss": 1.1221,
+      "step": 665
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0005,
+      "loss": 1.0857,
+      "step": 666
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0005,
+      "loss": 1.0857,
+      "step": 667
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0005,
+      "loss": 1.068,
+      "step": 668
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0005,
+      "loss": 1.1664,
+      "step": 669
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0005,
+      "loss": 1.1275,
+      "step": 670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0005,
+      "loss": 1.104,
+      "step": 671
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0005,
+      "loss": 1.0968,
+      "step": 672
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0005,
+      "loss": 1.0742,
+      "step": 673
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0005,
+      "loss": 1.1053,
+      "step": 674
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0005,
+      "loss": 1.098,
+      "step": 675
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0005,
+      "loss": 1.1067,
+      "step": 676
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0005,
+      "loss": 1.0613,
+      "step": 677
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0005,
+      "loss": 1.1308,
+      "step": 678
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0005,
+      "loss": 1.1962,
+      "step": 679
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0005,
+      "loss": 1.1398,
+      "step": 680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0005,
+      "loss": 1.1324,
+      "step": 681
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0005,
+      "loss": 1.0799,
+      "step": 682
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0005,
+      "loss": 1.0756,
+      "step": 683
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0005,
+      "loss": 1.0795,
+      "step": 684
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0005,
+      "loss": 1.0653,
+      "step": 685
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0005,
+      "loss": 1.1439,
+      "step": 686
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0005,
+      "loss": 1.0828,
+      "step": 687
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0005,
+      "loss": 1.0688,
+      "step": 688
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0005,
+      "loss": 1.1265,
+      "step": 689
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0005,
+      "loss": 1.0815,
+      "step": 690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0005,
+      "loss": 1.1125,
+      "step": 691
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0005,
+      "loss": 1.1091,
+      "step": 692
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0005,
+      "loss": 1.1174,
+      "step": 693
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0005,
+      "loss": 1.099,
+      "step": 694
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0005,
+      "loss": 1.1206,
+      "step": 695
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0005,
+      "loss": 1.0874,
+      "step": 696
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0005,
+      "loss": 1.1044,
+      "step": 697
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0005,
+      "loss": 1.0745,
+      "step": 698
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0005,
+      "loss": 1.0868,
+      "step": 699
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0005,
+      "loss": 1.1438,
+      "step": 700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0005,
+      "loss": 1.1195,
+      "step": 701
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0005,
+      "loss": 1.1014,
+      "step": 702
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0005,
+      "loss": 1.1201,
+      "step": 703
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0005,
+      "loss": 1.0637,
+      "step": 704
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0005,
+      "loss": 1.0999,
+      "step": 705
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0005,
+      "loss": 1.0813,
+      "step": 706
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0005,
+      "loss": 1.1071,
+      "step": 707
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0005,
+      "loss": 1.1175,
+      "step": 708
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0005,
+      "loss": 1.0857,
+      "step": 709
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0005,
+      "loss": 1.1159,
+      "step": 710
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0005,
+      "loss": 1.1585,
+      "step": 711
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0005,
+      "loss": 1.1417,
+      "step": 712
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0005,
+      "loss": 1.0656,
+      "step": 713
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0005,
+      "loss": 1.1289,
+      "step": 714
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0005,
+      "loss": 1.1022,
+      "step": 715
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0005,
+      "loss": 1.1085,
+      "step": 716
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0005,
+      "loss": 1.0929,
+      "step": 717
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0005,
+      "loss": 1.1263,
+      "step": 718
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0005,
+      "loss": 1.1048,
+      "step": 719
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0005,
+      "loss": 1.1229,
+      "step": 720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0005,
+      "loss": 1.0817,
+      "step": 721
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0005,
+      "loss": 1.1005,
+      "step": 722
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0005,
+      "loss": 1.0963,
+      "step": 723
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0005,
+      "loss": 1.1059,
+      "step": 724
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0005,
+      "loss": 1.0775,
+      "step": 725
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0005,
+      "loss": 1.0943,
+      "step": 726
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0005,
+      "loss": 1.132,
+      "step": 727
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0005,
+      "loss": 1.1033,
+      "step": 728
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0005,
+      "loss": 1.115,
+      "step": 729
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0005,
+      "loss": 1.1164,
+      "step": 730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0005,
+      "loss": 1.0839,
+      "step": 731
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0005,
+      "loss": 1.079,
+      "step": 732
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0005,
+      "loss": 1.1161,
+      "step": 733
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0005,
+      "loss": 1.0919,
+      "step": 734
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0005,
+      "loss": 1.1426,
+      "step": 735
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0005,
+      "loss": 1.1042,
+      "step": 736
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0005,
+      "loss": 1.1555,
+      "step": 737
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0005,
+      "loss": 1.0744,
+      "step": 738
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0005,
+      "loss": 1.0961,
+      "step": 739
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0005,
+      "loss": 1.087,
+      "step": 740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0005,
+      "loss": 1.1187,
+      "step": 741
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0005,
+      "loss": 1.0824,
+      "step": 742
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0005,
+      "loss": 1.1511,
+      "step": 743
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0005,
+      "loss": 1.0868,
+      "step": 744
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0005,
+      "loss": 1.1074,
+      "step": 745
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0005,
+      "loss": 1.0735,
+      "step": 746
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0005,
+      "loss": 1.0966,
+      "step": 747
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0005,
+      "loss": 1.1027,
+      "step": 748
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0005,
+      "loss": 1.1406,
+      "step": 749
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0005,
+      "loss": 1.1066,
+      "step": 750
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0005,
+      "loss": 1.1331,
+      "step": 751
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0005,
+      "loss": 1.1133,
+      "step": 752
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0005,
+      "loss": 1.0843,
+      "step": 753
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0005,
+      "loss": 1.1669,
+      "step": 754
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0005,
+      "loss": 1.1684,
+      "step": 755
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0005,
+      "loss": 1.1325,
+      "step": 756
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0005,
+      "loss": 1.1278,
+      "step": 757
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0005,
+      "loss": 1.1433,
+      "step": 758
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0005,
+      "loss": 1.1071,
+      "step": 759
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0005,
+      "loss": 1.0884,
+      "step": 760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0005,
+      "loss": 1.1432,
+      "step": 761
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0005,
+      "loss": 1.1135,
+      "step": 762
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0005,
+      "loss": 1.0831,
+      "step": 763
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0005,
+      "loss": 1.0794,
+      "step": 764
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0005,
+      "loss": 1.1232,
+      "step": 765
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0005,
+      "loss": 1.0984,
+      "step": 766
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0005,
+      "loss": 1.0881,
+      "step": 767
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0005,
+      "loss": 1.0875,
+      "step": 768
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0005,
+      "loss": 1.1447,
+      "step": 769
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0005,
+      "loss": 1.1168,
+      "step": 770
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0005,
+      "loss": 1.1088,
+      "step": 771
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0005,
+      "loss": 1.0764,
+      "step": 772
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0005,
+      "loss": 1.0941,
+      "step": 773
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0005,
+      "loss": 1.1003,
+      "step": 774
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0005,
+      "loss": 1.0956,
+      "step": 775
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0005,
+      "loss": 1.0576,
+      "step": 776
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0005,
+      "loss": 1.1346,
+      "step": 777
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0005,
+      "loss": 1.1333,
+      "step": 778
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0005,
+      "loss": 1.1416,
+      "step": 779
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0005,
+      "loss": 1.1313,
+      "step": 780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0005,
+      "loss": 1.0697,
+      "step": 781
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0005,
+      "loss": 1.1134,
+      "step": 782
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0005,
+      "loss": 1.1372,
+      "step": 783
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0005,
+      "loss": 1.1014,
+      "step": 784
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0005,
+      "loss": 1.0818,
+      "step": 785
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0005,
+      "loss": 1.123,
+      "step": 786
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0005,
+      "loss": 1.1136,
+      "step": 787
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0005,
+      "loss": 1.1025,
+      "step": 788
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0005,
+      "loss": 1.0854,
+      "step": 789
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0005,
+      "loss": 1.0855,
+      "step": 790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0005,
+      "loss": 1.0953,
+      "step": 791
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0005,
+      "loss": 1.1055,
+      "step": 792
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0005,
+      "loss": 1.1145,
+      "step": 793
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0005,
+      "loss": 1.0638,
+      "step": 794
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0005,
+      "loss": 1.1167,
+      "step": 795
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0005,
+      "loss": 1.099,
+      "step": 796
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0005,
+      "loss": 1.1113,
+      "step": 797
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0005,
+      "loss": 1.0363,
+      "step": 798
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0005,
+      "loss": 1.1093,
+      "step": 799
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0005,
+      "loss": 1.0849,
+      "step": 800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0005,
+      "loss": 1.1062,
+      "step": 801
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0005,
+      "loss": 1.0502,
+      "step": 802
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0005,
+      "loss": 1.1058,
+      "step": 803
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0005,
+      "loss": 1.1806,
+      "step": 804
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0005,
+      "loss": 1.1451,
+      "step": 805
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0005,
+      "loss": 1.0792,
+      "step": 806
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0005,
+      "loss": 1.0959,
+      "step": 807
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0005,
+      "loss": 1.1004,
+      "step": 808
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0005,
+      "loss": 1.1143,
+      "step": 809
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0005,
+      "loss": 1.1294,
+      "step": 810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0005,
+      "loss": 1.1008,
+      "step": 811
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0005,
+      "loss": 1.1182,
+      "step": 812
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0005,
+      "loss": 1.117,
+      "step": 813
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0005,
+      "loss": 1.1207,
+      "step": 814
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0005,
+      "loss": 1.1213,
+      "step": 815
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0005,
+      "loss": 1.0676,
+      "step": 816
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0005,
+      "loss": 1.1163,
+      "step": 817
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0005,
+      "loss": 1.0947,
+      "step": 818
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0005,
+      "loss": 1.1144,
+      "step": 819
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0005,
+      "loss": 1.0878,
+      "step": 820
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0005,
+      "loss": 1.1174,
+      "step": 821
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0005,
+      "loss": 1.0926,
+      "step": 822
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0005,
+      "loss": 1.1066,
+      "step": 823
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0005,
+      "loss": 1.1115,
+      "step": 824
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0005,
+      "loss": 1.0585,
+      "step": 825
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0005,
+      "loss": 1.1201,
+      "step": 826
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0005,
+      "loss": 1.0767,
+      "step": 827
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0005,
+      "loss": 1.0875,
+      "step": 828
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0005,
+      "loss": 1.0978,
+      "step": 829
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0005,
+      "loss": 1.1338,
+      "step": 830
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0005,
+      "loss": 1.108,
+      "step": 831
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005,
+      "loss": 1.0795,
+      "step": 832
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005,
+      "loss": 1.11,
+      "step": 833
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005,
+      "loss": 1.0496,
+      "step": 834
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005,
+      "loss": 1.0698,
+      "step": 835
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005,
+      "loss": 1.099,
+      "step": 836
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005,
+      "loss": 1.1051,
+      "step": 837
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005,
+      "loss": 1.0613,
+      "step": 838
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005,
+      "loss": 1.116,
+      "step": 839
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005,
+      "loss": 1.0896,
+      "step": 840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005,
+      "loss": 1.1035,
+      "step": 841
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005,
+      "loss": 1.0452,
+      "step": 842
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005,
+      "loss": 1.0899,
+      "step": 843
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005,
+      "loss": 1.1036,
+      "step": 844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005,
+      "loss": 1.1196,
+      "step": 845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005,
+      "loss": 1.0674,
+      "step": 846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005,
+      "loss": 1.0857,
+      "step": 847
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005,
+      "loss": 1.0712,
+      "step": 848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005,
+      "loss": 1.1294,
+      "step": 849
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005,
+      "loss": 1.0668,
+      "step": 850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005,
+      "loss": 1.0598,
+      "step": 851
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005,
+      "loss": 1.1109,
+      "step": 852
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005,
+      "loss": 1.0763,
+      "step": 853
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005,
+      "loss": 1.1044,
+      "step": 854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005,
+      "loss": 1.1576,
+      "step": 855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005,
+      "loss": 1.1349,
+      "step": 856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005,
+      "loss": 1.1037,
+      "step": 857
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005,
+      "loss": 1.1247,
+      "step": 858
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005,
+      "loss": 1.1302,
+      "step": 859
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005,
+      "loss": 1.1072,
+      "step": 860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005,
+      "loss": 1.0663,
+      "step": 861
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005,
+      "loss": 1.119,
+      "step": 862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005,
+      "loss": 1.1027,
+      "step": 863
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005,
+      "loss": 1.1633,
+      "step": 864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005,
+      "loss": 1.0857,
+      "step": 865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005,
+      "loss": 1.1031,
+      "step": 866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005,
+      "loss": 1.0759,
+      "step": 867
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005,
+      "loss": 1.0975,
+      "step": 868
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005,
+      "loss": 1.1353,
+      "step": 869
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005,
+      "loss": 1.0713,
+      "step": 870
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005,
+      "loss": 1.1092,
+      "step": 871
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005,
+      "loss": 1.1258,
+      "step": 872
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005,
+      "loss": 1.0857,
+      "step": 873
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005,
+      "loss": 1.0982,
+      "step": 874
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005,
+      "loss": 1.0857,
+      "step": 875
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005,
+      "loss": 1.047,
+      "step": 876
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005,
+      "loss": 1.0826,
+      "step": 877
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005,
+      "loss": 1.0755,
+      "step": 878
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005,
+      "loss": 1.1544,
+      "step": 879
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005,
+      "loss": 1.1101,
+      "step": 880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005,
+      "loss": 1.0808,
+      "step": 881
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005,
+      "loss": 1.0906,
+      "step": 882
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005,
+      "loss": 1.0888,
+      "step": 883
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005,
+      "loss": 1.0788,
+      "step": 884
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005,
+      "loss": 1.0749,
+      "step": 885
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005,
+      "loss": 1.0795,
+      "step": 886
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005,
+      "loss": 1.0649,
+      "step": 887
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005,
+      "loss": 1.067,
+      "step": 888
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005,
+      "loss": 1.0922,
+      "step": 889
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005,
+      "loss": 1.1163,
+      "step": 890
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005,
+      "loss": 1.0979,
+      "step": 891
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005,
+      "loss": 1.0719,
+      "step": 892
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005,
+      "loss": 1.0796,
+      "step": 893
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005,
+      "loss": 1.1005,
+      "step": 894
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005,
+      "loss": 1.0766,
+      "step": 895
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005,
+      "loss": 1.1292,
+      "step": 896
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005,
+      "loss": 1.1281,
+      "step": 897
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005,
+      "loss": 1.0687,
+      "step": 898
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005,
+      "loss": 1.0858,
+      "step": 899
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005,
+      "loss": 1.0695,
+      "step": 900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005,
+      "loss": 1.0969,
+      "step": 901
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005,
+      "loss": 1.0681,
+      "step": 902
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005,
+      "loss": 1.1084,
+      "step": 903
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005,
+      "loss": 1.0828,
+      "step": 904
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005,
+      "loss": 1.1158,
+      "step": 905
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005,
+      "loss": 1.1029,
+      "step": 906
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005,
+      "loss": 1.0668,
+      "step": 907
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005,
+      "loss": 1.0865,
+      "step": 908
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005,
+      "loss": 1.0623,
+      "step": 909
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005,
+      "loss": 1.0685,
+      "step": 910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005,
+      "loss": 1.144,
+      "step": 911
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005,
+      "loss": 1.1082,
+      "step": 912
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005,
+      "loss": 1.1145,
+      "step": 913
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005,
+      "loss": 1.1184,
+      "step": 914
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005,
+      "loss": 1.0921,
+      "step": 915
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005,
+      "loss": 1.0661,
+      "step": 916
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005,
+      "loss": 1.0955,
+      "step": 917
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005,
+      "loss": 1.114,
+      "step": 918
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005,
+      "loss": 1.2113,
+      "step": 919
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 6000,
+  "num_train_epochs": 7,
+  "save_steps": 500,
+  "total_flos": 2.320438859784192e+17,
+  "trial_name": null,
+  "trial_params": null
+}