diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,56286 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 9376,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.666666666666667e-07,
+      "loss": 1.127,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 1.4115,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.1369,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 1.209,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.1893,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.1133,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.666666666666667e-06,
+      "loss": 1.495,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 1.3286,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-06,
+      "loss": 1.4895,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.5401,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.333333333333334e-06,
+      "loss": 1.9378,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.1589,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.666666666666668e-06,
+      "loss": 1.3567,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 1.3356,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 1.5226,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 1.1813,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1333333333333334e-05,
+      "loss": 1.2342,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 1.7119,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2666666666666668e-05,
+      "loss": 1.2515,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.27,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 1.624,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4666666666666668e-05,
+      "loss": 1.1865,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5333333333333334e-05,
+      "loss": 1.6761,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.3476,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 1.4337,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 1.4853,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8e-05,
+      "loss": 1.4205,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 1.2968,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9333333333333333e-05,
+      "loss": 1.5865,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 1.5479,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0666666666666666e-05,
+      "loss": 1.3665,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1333333333333335e-05,
+      "loss": 1.22,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 1.4975,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2666666666666668e-05,
+      "loss": 1.3368,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3333333333333336e-05,
+      "loss": 1.3579,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4e-05,
+      "loss": 1.4886,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.466666666666667e-05,
+      "loss": 1.4239,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5333333333333337e-05,
+      "loss": 1.5366,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 1.7977,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 1.4509,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.733333333333333e-05,
+      "loss": 1.5004,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 1.2001,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8666666666666668e-05,
+      "loss": 1.3609,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9333333333333336e-05,
+      "loss": 1.4067,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-05,
+      "loss": 1.4726,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.066666666666667e-05,
+      "loss": 1.5731,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1333333333333334e-05,
+      "loss": 1.5043,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 1.3045,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.266666666666667e-05,
+      "loss": 1.0624,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 1.3541,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 1.2885,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.466666666666667e-05,
+      "loss": 1.4096,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5333333333333336e-05,
+      "loss": 1.3576,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-05,
+      "loss": 1.3608,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6666666666666666e-05,
+      "loss": 1.2712,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.733333333333334e-05,
+      "loss": 1.3711,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8e-05,
+      "loss": 1.3113,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.866666666666667e-05,
+      "loss": 1.2776,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.933333333333333e-05,
+      "loss": 1.4833,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4e-05,
+      "loss": 1.401,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.066666666666667e-05,
+      "loss": 1.2806,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.133333333333333e-05,
+      "loss": 1.3004,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2e-05,
+      "loss": 1.3411,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.266666666666667e-05,
+      "loss": 1.1288,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3333333333333334e-05,
+      "loss": 1.8089,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 1.4489,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.466666666666667e-05,
+      "loss": 1.5561,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5333333333333335e-05,
+      "loss": 1.2847,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 1.6469,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.666666666666667e-05,
+      "loss": 1.4525,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7333333333333336e-05,
+      "loss": 1.7741,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8e-05,
+      "loss": 1.4905,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.866666666666667e-05,
+      "loss": 1.5127,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.933333333333334e-05,
+      "loss": 1.278,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-05,
+      "loss": 1.1986,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.0666666666666674e-05,
+      "loss": 1.2931,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.133333333333333e-05,
+      "loss": 1.502,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 1.3635,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.266666666666666e-05,
+      "loss": 1.6804,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.333333333333333e-05,
+      "loss": 1.5403,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4000000000000005e-05,
+      "loss": 1.436,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.466666666666666e-05,
+      "loss": 1.6303,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5333333333333334e-05,
+      "loss": 1.2198,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 1.8833,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.666666666666667e-05,
+      "loss": 1.1595,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.7333333333333336e-05,
+      "loss": 1.5795,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8e-05,
+      "loss": 1.3304,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.866666666666667e-05,
+      "loss": 1.3615,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9333333333333343e-05,
+      "loss": 1.1705,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6e-05,
+      "loss": 1.7922,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.066666666666667e-05,
+      "loss": 1.2268,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.133333333333334e-05,
+      "loss": 1.3521,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.2e-05,
+      "loss": 1.2906,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.266666666666667e-05,
+      "loss": 1.3189,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.333333333333333e-05,
+      "loss": 1.446,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 1.3511,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.466666666666666e-05,
+      "loss": 1.4333,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.533333333333334e-05,
+      "loss": 1.289,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.6e-05,
+      "loss": 1.2087,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 1.375,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.733333333333333e-05,
+      "loss": 1.2301,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 1.3112,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.866666666666666e-05,
+      "loss": 1.4281,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.933333333333334e-05,
+      "loss": 1.3309,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7e-05,
+      "loss": 1.3861,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.066666666666667e-05,
+      "loss": 1.2807,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.133333333333334e-05,
+      "loss": 1.4146,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.2e-05,
+      "loss": 1.2647,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.266666666666667e-05,
+      "loss": 1.111,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.333333333333333e-05,
+      "loss": 1.6217,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.4e-05,
+      "loss": 1.3323,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.466666666666667e-05,
+      "loss": 1.6769,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.533333333333334e-05,
+      "loss": 1.2457,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.6e-05,
+      "loss": 1.5833,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.666666666666667e-05,
+      "loss": 1.3401,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.733333333333333e-05,
+      "loss": 1.6458,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 1.2087,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.866666666666666e-05,
+      "loss": 1.2158,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.933333333333334e-05,
+      "loss": 1.3255,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8e-05,
+      "loss": 1.2276,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.066666666666667e-05,
+      "loss": 1.1696,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.133333333333334e-05,
+      "loss": 1.4312,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.2e-05,
+      "loss": 1.254,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.266666666666667e-05,
+      "loss": 1.2986,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.333333333333334e-05,
+      "loss": 1.7244,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.4e-05,
+      "loss": 1.5303,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.466666666666667e-05,
+      "loss": 1.5499,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.533333333333334e-05,
+      "loss": 1.4238,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.6e-05,
+      "loss": 1.5946,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.666666666666667e-05,
+      "loss": 1.0713,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.733333333333333e-05,
+      "loss": 1.1396,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 1.27,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.866666666666668e-05,
+      "loss": 1.3755,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.933333333333334e-05,
+      "loss": 1.8665,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9e-05,
+      "loss": 1.4668,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.066666666666667e-05,
+      "loss": 1.6113,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.133333333333334e-05,
+      "loss": 1.3567,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 1.1649,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.266666666666666e-05,
+      "loss": 1.1624,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.333333333333334e-05,
+      "loss": 1.3018,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.4e-05,
+      "loss": 1.2171,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.466666666666667e-05,
+      "loss": 1.1493,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.533333333333334e-05,
+      "loss": 1.3024,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.6e-05,
+      "loss": 1.6234,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.666666666666667e-05,
+      "loss": 1.1422,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.733333333333335e-05,
+      "loss": 1.6167,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.8e-05,
+      "loss": 1.2896,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.866666666666668e-05,
+      "loss": 1.5833,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.933333333333334e-05,
+      "loss": 1.7162,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001,
+      "loss": 1.4108,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010066666666666667,
+      "loss": 1.2992,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010133333333333335,
+      "loss": 1.3092,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010200000000000001,
+      "loss": 1.4397,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010266666666666666,
+      "loss": 1.3373,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010333333333333334,
+      "loss": 1.5093,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010400000000000001,
+      "loss": 1.6736,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010466666666666667,
+      "loss": 1.5298,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010533333333333332,
+      "loss": 1.3274,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010600000000000002,
+      "loss": 1.2479,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010666666666666667,
+      "loss": 1.1106,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010733333333333333,
+      "loss": 1.4663,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010800000000000001,
+      "loss": 1.1891,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010866666666666667,
+      "loss": 1.3444,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00010933333333333333,
+      "loss": 1.6852,
+      "step": 164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011000000000000002,
+      "loss": 1.5317,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011066666666666667,
+      "loss": 1.276,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011133333333333333,
+      "loss": 1.4352,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011200000000000001,
+      "loss": 1.3292,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011266666666666668,
+      "loss": 1.2348,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011333333333333334,
+      "loss": 1.637,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 1.5105,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011466666666666667,
+      "loss": 1.3253,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011533333333333334,
+      "loss": 1.4597,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000116,
+      "loss": 1.3436,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011666666666666668,
+      "loss": 1.4006,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011733333333333334,
+      "loss": 1.1943,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000118,
+      "loss": 1.2946,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011866666666666669,
+      "loss": 1.1453,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00011933333333333334,
+      "loss": 1.2077,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00012,
+      "loss": 1.4301,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00012066666666666668,
+      "loss": 1.5338,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00012133333333333335,
+      "loss": 1.6929,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000122,
+      "loss": 1.2452,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00012266666666666668,
+      "loss": 1.4826,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00012333333333333334,
+      "loss": 1.4121,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000124,
+      "loss": 1.3574,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00012466666666666667,
+      "loss": 1.7534,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00012533333333333334,
+      "loss": 1.1813,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000126,
+      "loss": 1.5632,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00012666666666666666,
+      "loss": 1.3045,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00012733333333333336,
+      "loss": 1.302,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00012800000000000002,
+      "loss": 1.4596,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00012866666666666666,
+      "loss": 1.0887,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00012933333333333332,
+      "loss": 1.3458,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013000000000000002,
+      "loss": 1.2491,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013066666666666668,
+      "loss": 1.5106,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013133333333333332,
+      "loss": 1.286,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000132,
+      "loss": 1.6784,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013266666666666667,
+      "loss": 1.1746,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 1.1452,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000134,
+      "loss": 1.2427,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013466666666666667,
+      "loss": 1.1736,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013533333333333333,
+      "loss": 1.4271,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013600000000000003,
+      "loss": 1.1216,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013666666666666666,
+      "loss": 1.3565,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013733333333333333,
+      "loss": 1.6959,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000138,
+      "loss": 1.4487,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013866666666666669,
+      "loss": 1.4518,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013933333333333335,
+      "loss": 1.0676,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00014,
+      "loss": 1.4076,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00014066666666666668,
+      "loss": 1.2602,
+      "step": 211
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00014133333333333334,
+      "loss": 1.483,
+      "step": 212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000142,
+      "loss": 1.2141,
+      "step": 213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00014266666666666667,
+      "loss": 1.3684,
+      "step": 214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00014333333333333334,
+      "loss": 1.1309,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000144,
+      "loss": 1.8714,
+      "step": 216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001446666666666667,
+      "loss": 1.2245,
+      "step": 217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00014533333333333333,
+      "loss": 1.2858,
+      "step": 218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000146,
+      "loss": 1.336,
+      "step": 219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00014666666666666666,
+      "loss": 1.191,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00014733333333333335,
+      "loss": 1.3248,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000148,
+      "loss": 1.3838,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00014866666666666666,
+      "loss": 0.9002,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00014933333333333335,
+      "loss": 1.4914,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 1.3025,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00015066666666666668,
+      "loss": 1.431,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00015133333333333334,
+      "loss": 1.523,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000152,
+      "loss": 1.256,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00015266666666666667,
+      "loss": 1.4039,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00015333333333333334,
+      "loss": 1.361,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000154,
+      "loss": 1.0009,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00015466666666666667,
+      "loss": 1.3906,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00015533333333333333,
+      "loss": 1.3764,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00015600000000000002,
+      "loss": 1.2491,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00015666666666666666,
+      "loss": 1.0725,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00015733333333333333,
+      "loss": 1.6447,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00015800000000000002,
+      "loss": 1.3279,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00015866666666666668,
+      "loss": 1.5658,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00015933333333333332,
+      "loss": 1.0235,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00016,
+      "loss": 1.7763,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00016066666666666668,
+      "loss": 1.1093,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00016133333333333334,
+      "loss": 1.2677,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000162,
+      "loss": 1.5862,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00016266666666666667,
+      "loss": 1.4258,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00016333333333333334,
+      "loss": 1.3567,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000164,
+      "loss": 1.2566,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00016466666666666667,
+      "loss": 1.4611,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00016533333333333333,
+      "loss": 1.3054,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000166,
+      "loss": 1.3394,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001666666666666667,
+      "loss": 1.4737,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00016733333333333335,
+      "loss": 1.3701,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000168,
+      "loss": 1.2219,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00016866666666666668,
+      "loss": 1.3412,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00016933333333333335,
+      "loss": 1.4314,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017,
+      "loss": 1.5236,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017066666666666668,
+      "loss": 1.517,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017133333333333334,
+      "loss": 1.1889,
+      "step": 257
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000172,
+      "loss": 1.3229,
+      "step": 258
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017266666666666667,
+      "loss": 1.2269,
+      "step": 259
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017333333333333334,
+      "loss": 1.2934,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000174,
+      "loss": 1.6912,
+      "step": 261
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017466666666666667,
+      "loss": 1.2405,
+      "step": 262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017533333333333336,
+      "loss": 1.6569,
+      "step": 263
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017600000000000002,
+      "loss": 1.3825,
+      "step": 264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017666666666666666,
+      "loss": 1.3289,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017733333333333335,
+      "loss": 1.3035,
+      "step": 266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017800000000000002,
+      "loss": 1.6307,
+      "step": 267
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017866666666666668,
+      "loss": 1.6749,
+      "step": 268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00017933333333333332,
+      "loss": 1.3711,
+      "step": 269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00018,
+      "loss": 1.2486,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00018066666666666668,
+      "loss": 1.4425,
+      "step": 271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00018133333333333334,
+      "loss": 1.1891,
+      "step": 272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000182,
+      "loss": 1.7831,
+      "step": 273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00018266666666666667,
+      "loss": 1.3578,
+      "step": 274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00018333333333333334,
+      "loss": 1.3151,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00018400000000000003,
+      "loss": 1.0363,
+      "step": 276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00018466666666666666,
+      "loss": 1.2137,
+      "step": 277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00018533333333333333,
+      "loss": 1.6683,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00018600000000000002,
+      "loss": 1.3256,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001866666666666667,
+      "loss": 1.4339,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00018733333333333335,
+      "loss": 1.2215,
+      "step": 281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000188,
+      "loss": 1.4178,
+      "step": 282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00018866666666666668,
+      "loss": 1.0471,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00018933333333333335,
+      "loss": 1.4747,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019,
+      "loss": 1.2473,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019066666666666668,
+      "loss": 1.6723,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019133333333333334,
+      "loss": 1.3077,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000192,
+      "loss": 1.4837,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001926666666666667,
+      "loss": 1.1452,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019333333333333333,
+      "loss": 1.3503,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000194,
+      "loss": 1.8228,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001946666666666667,
+      "loss": 1.5872,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019533333333333336,
+      "loss": 1.0852,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000196,
+      "loss": 1.1119,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019666666666666666,
+      "loss": 1.2665,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019733333333333335,
+      "loss": 1.1409,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019800000000000002,
+      "loss": 1.0668,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019866666666666668,
+      "loss": 1.2034,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019933333333333334,
+      "loss": 1.8217,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 1.3082,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999999940092559,
+      "loss": 1.4116,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999999760370242,
+      "loss": 1.416,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999994608330712,
+      "loss": 1.3036,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019999990414810826,
+      "loss": 1.3962,
+      "step": 304
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999985023143262,
+      "loss": 1.411,
+      "step": 305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999978433328665,
+      "loss": 1.2147,
+      "step": 306
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999970645367827,
+      "loss": 1.3136,
+      "step": 307
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001999996165926168,
+      "loss": 1.4999,
+      "step": 308
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999951475011297,
+      "loss": 1.0482,
+      "step": 309
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999940092617905,
+      "loss": 1.4527,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999927512082865,
+      "loss": 1.3298,
+      "step": 311
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999913733407686,
+      "loss": 1.3078,
+      "step": 312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999898756594014,
+      "loss": 2.0534,
+      "step": 313
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999882581643643,
+      "loss": 1.642,
+      "step": 314
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001999986520855852,
+      "loss": 1.5113,
+      "step": 315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999846637340718,
+      "loss": 1.6165,
+      "step": 316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999826867992467,
+      "loss": 1.55,
+      "step": 317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999805900516128,
+      "loss": 1.7681,
+      "step": 318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999783734914224,
+      "loss": 1.2431,
+      "step": 319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999760371189405,
+      "loss": 1.1919,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001999973580934447,
+      "loss": 1.4577,
+      "step": 321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999710049382362,
+      "loss": 1.4623,
+      "step": 322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999683091306166,
+      "loss": 0.9554,
+      "step": 323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001999965493511912,
+      "loss": 1.1917,
+      "step": 324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999625580824587,
+      "loss": 1.3347,
+      "step": 325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999595028426093,
+      "loss": 1.2061,
+      "step": 326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999563277927292,
+      "loss": 1.5251,
+      "step": 327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999530329331991,
+      "loss": 1.3487,
+      "step": 328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999496182644135,
+      "loss": 1.2628,
+      "step": 329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999460837867822,
+      "loss": 1.6685,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001999942429500728,
+      "loss": 1.3303,
+      "step": 331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999386554066888,
+      "loss": 1.2922,
+      "step": 332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999347615051169,
+      "loss": 1.7448,
+      "step": 333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999307477964795,
+      "loss": 1.7123,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999266142812564,
+      "loss": 1.3307,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999223609599436,
+      "loss": 1.3606,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999179878330505,
+      "loss": 1.6269,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001999913494901101,
+      "loss": 1.1301,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999088821646334,
+      "loss": 1.4347,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019999041496242007,
+      "loss": 1.4579,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019998992972803693,
+      "loss": 1.1457,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019998943251337212,
+      "loss": 1.2137,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001999889233184852,
+      "loss": 1.3197,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019998840214343716,
+      "loss": 1.2415,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019998786898829045,
+      "loss": 1.4917,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019998732385310896,
+      "loss": 1.0129,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000199986766737958,
+      "loss": 1.5587,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001999861976429043,
+      "loss": 1.3264,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019998561656801606,
+      "loss": 1.4904,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019998502351336296,
+      "loss": 1.0977,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019998441847901595,
+      "loss": 1.2308,
+      "step": 351
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001999838014650476,
+      "loss": 1.039,
+      "step": 352
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001999831724715318,
+      "loss": 1.087,
+      "step": 353
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019998253149854396,
+      "loss": 1.4565,
+      "step": 354
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019998187854616082,
+      "loss": 1.4997,
+      "step": 355
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019998121361446063,
+      "loss": 1.353,
+      "step": 356
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019998053670352306,
+      "loss": 1.6063,
+      "step": 357
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019997984781342925,
+      "loss": 1.4761,
+      "step": 358
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001999791469442617,
+      "loss": 1.5571,
+      "step": 359
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019997843409610437,
+      "loss": 1.3749,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019997770926904272,
+      "loss": 1.7424,
+      "step": 361
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019997697246316355,
+      "loss": 1.3897,
+      "step": 362
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019997622367855518,
+      "loss": 1.2946,
+      "step": 363
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001999754629153073,
+      "loss": 1.1622,
+      "step": 364
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019997469017351104,
+      "loss": 1.6455,
+      "step": 365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019997390545325905,
+      "loss": 1.1702,
+      "step": 366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001999731087546453,
+      "loss": 1.1692,
+      "step": 367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019997230007776524,
+      "loss": 1.4354,
+      "step": 368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001999714794227158,
+      "loss": 1.1376,
+      "step": 369
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001999706467895953,
+      "loss": 1.4793,
+      "step": 370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019996980217850348,
+      "loss": 1.5344,
+      "step": 371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001999689455895415,
+      "loss": 1.417,
+      "step": 372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001999680770228121,
+      "loss": 1.3982,
+      "step": 373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019996719647841926,
+      "loss": 1.5749,
+      "step": 374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001999663039564685,
+      "loss": 1.4526,
+      "step": 375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001999653994570668,
+      "loss": 1.2305,
+      "step": 376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019996448298032244,
+      "loss": 1.2964,
+      "step": 377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019996355452634533,
+      "loss": 1.549,
+      "step": 378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001999626140952467,
+      "loss": 1.8852,
+      "step": 379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019996166168713913,
+      "loss": 1.5208,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001999606973021368,
+      "loss": 1.3924,
+      "step": 381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019995972094035528,
+      "loss": 1.2175,
+      "step": 382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019995873260191153,
+      "loss": 1.4629,
+      "step": 383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019995773228692395,
+      "loss": 1.1534,
+      "step": 384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019995671999551245,
+      "loss": 1.2569,
+      "step": 385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019995569572779824,
+      "loss": 1.5535,
+      "step": 386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019995465948390408,
+      "loss": 1.46,
+      "step": 387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019995361126395414,
+      "loss": 1.6316,
+      "step": 388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019995255106807397,
+      "loss": 1.2402,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019995147889639067,
+      "loss": 1.1784,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019995039474903262,
+      "loss": 1.5168,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019994929862612976,
+      "loss": 1.3683,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019994819052781347,
+      "loss": 1.7636,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001999470704542164,
+      "loss": 1.6235,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019994593840547283,
+      "loss": 1.7287,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019994479438171841,
+      "loss": 1.5131,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019994363838309016,
+      "loss": 1.4888,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001999424704097266,
+      "loss": 1.4607,
+      "step": 398
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019994129046176768,
+      "loss": 1.2721,
+      "step": 399
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019994009853935478,
+      "loss": 1.0406,
+      "step": 400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999388946426307,
+      "loss": 1.242,
+      "step": 401
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999376787717397,
+      "loss": 1.1578,
+      "step": 402
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019993645092682746,
+      "loss": 1.4219,
+      "step": 403
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019993521110804109,
+      "loss": 1.4354,
+      "step": 404
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019993395931552907,
+      "loss": 1.3002,
+      "step": 405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999326955494415,
+      "loss": 1.3309,
+      "step": 406
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019993141980992972,
+      "loss": 1.1561,
+      "step": 407
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019993013209714658,
+      "loss": 1.1859,
+      "step": 408
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019992883241124645,
+      "loss": 1.6529,
+      "step": 409
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019992752075238498,
+      "loss": 1.3124,
+      "step": 410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999261971207193,
+      "loss": 1.696,
+      "step": 411
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019992486151640808,
+      "loss": 1.7675,
+      "step": 412
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019992351393961132,
+      "loss": 0.9653,
+      "step": 413
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999221543904904,
+      "loss": 0.877,
+      "step": 414
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019992078286920834,
+      "loss": 1.328,
+      "step": 415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019991939937592942,
+      "loss": 1.2595,
+      "step": 416
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019991800391081933,
+      "loss": 1.4768,
+      "step": 417
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019991659647404538,
+      "loss": 1.4767,
+      "step": 418
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999151770657761,
+      "loss": 1.4308,
+      "step": 419
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019991374568618166,
+      "loss": 1.7027,
+      "step": 420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999123023354335,
+      "loss": 1.3152,
+      "step": 421
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019991084701370454,
+      "loss": 1.3138,
+      "step": 422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999093797211692,
+      "loss": 1.4552,
+      "step": 423
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019990790045800325,
+      "loss": 1.2847,
+      "step": 424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019990640922438392,
+      "loss": 0.9637,
+      "step": 425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999049060204899,
+      "loss": 1.3576,
+      "step": 426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999033908465013,
+      "loss": 0.9402,
+      "step": 427
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019990186370259964,
+      "loss": 1.5834,
+      "step": 428
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001999003245889679,
+      "loss": 1.5707,
+      "step": 429
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001998987735057905,
+      "loss": 1.5802,
+      "step": 430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001998972104532533,
+      "loss": 0.773,
+      "step": 431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019989563543154355,
+      "loss": 1.7507,
+      "step": 432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019989404844084993,
+      "loss": 1.4836,
+      "step": 433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019989244948136268,
+      "loss": 1.2847,
+      "step": 434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019989083855327327,
+      "loss": 1.2731,
+      "step": 435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001998892156567748,
+      "loss": 1.6513,
+      "step": 436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019988758079206168,
+      "loss": 1.381,
+      "step": 437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019988593395932976,
+      "loss": 1.2468,
+      "step": 438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019988427515877638,
+      "loss": 1.2622,
+      "step": 439
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019988260439060033,
+      "loss": 1.1616,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019988092165500177,
+      "loss": 1.5562,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019987922695218226,
+      "loss": 1.5719,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019987752028234493,
+      "loss": 1.563,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001998758016456942,
+      "loss": 1.2208,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019987407104243604,
+      "loss": 1.2008,
+      "step": 445
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019987232847277779,
+      "loss": 1.37,
+      "step": 446
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019987057393692823,
+      "loss": 1.1133,
+      "step": 447
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019986880743509756,
+      "loss": 1.1362,
+      "step": 448
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019986702896749746,
+      "loss": 1.4561,
+      "step": 449
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000199865238534341,
+      "loss": 1.3835,
+      "step": 450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001998634361358427,
+      "loss": 1.5041,
+      "step": 451
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019986162177221855,
+      "loss": 1.1814,
+      "step": 452
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019985979544368587,
+      "loss": 1.4526,
+      "step": 453
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019985795715046355,
+      "loss": 1.3821,
+      "step": 454
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019985610689277178,
+      "loss": 1.5113,
+      "step": 455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019985424467083233,
+      "loss": 1.2497,
+      "step": 456
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019985237048486824,
+      "loss": 1.5954,
+      "step": 457
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019985048433510412,
+      "loss": 1.4213,
+      "step": 458
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019984858622176593,
+      "loss": 1.2254,
+      "step": 459
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001998466761450811,
+      "loss": 1.3499,
+      "step": 460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019984475410527849,
+      "loss": 1.2084,
+      "step": 461
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019984282010258842,
+      "loss": 1.1953,
+      "step": 462
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019984087413724253,
+      "loss": 1.337,
+      "step": 463
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019983891620947405,
+      "loss": 1.3365,
+      "step": 464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019983694631951751,
+      "loss": 1.43,
+      "step": 465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019983496446760902,
+      "loss": 0.9646,
+      "step": 466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019983297065398595,
+      "loss": 1.4331,
+      "step": 467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001998309648788872,
+      "loss": 1.2251,
+      "step": 468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019982894714255315,
+      "loss": 1.318,
+      "step": 469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019982691744522548,
+      "loss": 1.9402,
+      "step": 470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019982487578714742,
+      "loss": 1.2525,
+      "step": 471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019982282216856362,
+      "loss": 1.2159,
+      "step": 472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019982075658972007,
+      "loss": 1.3393,
+      "step": 473
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019981867905086428,
+      "loss": 1.4129,
+      "step": 474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001998165895522452,
+      "loss": 1.5442,
+      "step": 475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019981448809411315,
+      "loss": 1.6061,
+      "step": 476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001998123746767199,
+      "loss": 1.5108,
+      "step": 477
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019981024930031872,
+      "loss": 1.4287,
+      "step": 478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019980811196516424,
+      "loss": 1.5375,
+      "step": 479
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019980596267151252,
+      "loss": 1.5325,
+      "step": 480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019980380141962112,
+      "loss": 1.3712,
+      "step": 481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019980162820974897,
+      "loss": 1.3338,
+      "step": 482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019979944304215644,
+      "loss": 1.3343,
+      "step": 483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019979724591710535,
+      "loss": 1.2791,
+      "step": 484
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019979503683485893,
+      "loss": 1.1656,
+      "step": 485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019979281579568192,
+      "loss": 1.5646,
+      "step": 486
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001997905827998404,
+      "loss": 1.2297,
+      "step": 487
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019978833784760192,
+      "loss": 1.1471,
+      "step": 488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019978608093923543,
+      "loss": 1.2017,
+      "step": 489
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019978381207501138,
+      "loss": 1.3429,
+      "step": 490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001997815312552016,
+      "loss": 1.4504,
+      "step": 491
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019977923848007935,
+      "loss": 1.1203,
+      "step": 492
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019977693374991938,
+      "loss": 1.4337,
+      "step": 493
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019977461706499776,
+      "loss": 1.4329,
+      "step": 494
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019977228842559214,
+      "loss": 1.5728,
+      "step": 495
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001997699478319815,
+      "loss": 1.133,
+      "step": 496
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019976759528444628,
+      "loss": 1.1934,
+      "step": 497
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001997652307832683,
+      "loss": 1.338,
+      "step": 498
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019976285432873092,
+      "loss": 1.4318,
+      "step": 499
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019976046592111888,
+      "loss": 1.1005,
+      "step": 500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001997580655607183,
+      "loss": 0.9359,
+      "step": 501
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019975565324781682,
+      "loss": 1.1487,
+      "step": 502
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019975322898270345,
+      "loss": 1.3462,
+      "step": 503
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019975079276566863,
+      "loss": 1.3802,
+      "step": 504
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019974834459700433,
+      "loss": 1.4368,
+      "step": 505
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001997458844770038,
+      "loss": 1.5215,
+      "step": 506
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019974341240596184,
+      "loss": 1.3325,
+      "step": 507
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019974092838417464,
+      "loss": 1.3816,
+      "step": 508
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019973843241193978,
+      "loss": 1.2996,
+      "step": 509
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001997359244895564,
+      "loss": 1.2227,
+      "step": 510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001997334046173249,
+      "loss": 1.1702,
+      "step": 511
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019973087279554722,
+      "loss": 1.2962,
+      "step": 512
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019972832902452676,
+      "loss": 1.4378,
+      "step": 513
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019972577330456824,
+      "loss": 1.0434,
+      "step": 514
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001997232056359779,
+      "loss": 1.174,
+      "step": 515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019972062601906338,
+      "loss": 1.1234,
+      "step": 516
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019971803445413376,
+      "loss": 1.4487,
+      "step": 517
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019971543094149955,
+      "loss": 1.2453,
+      "step": 518
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001997128154814727,
+      "loss": 1.8258,
+      "step": 519
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019971018807436654,
+      "loss": 1.6672,
+      "step": 520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019970754872049594,
+      "loss": 1.5341,
+      "step": 521
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019970489742017703,
+      "loss": 1.1632,
+      "step": 522
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001997022341737276,
+      "loss": 1.3643,
+      "step": 523
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019969955898146665,
+      "loss": 1.3722,
+      "step": 524
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019969687184371476,
+      "loss": 1.3551,
+      "step": 525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019969417276079386,
+      "loss": 1.5183,
+      "step": 526
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019969146173302735,
+      "loss": 1.6111,
+      "step": 527
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019968873876074006,
+      "loss": 1.5808,
+      "step": 528
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019968600384425824,
+      "loss": 1.1856,
+      "step": 529
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019968325698390958,
+      "loss": 1.0864,
+      "step": 530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019968049818002315,
+      "loss": 1.3313,
+      "step": 531
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019967772743292956,
+      "loss": 1.5357,
+      "step": 532
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019967494474296076,
+      "loss": 1.4401,
+      "step": 533
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019967215011045016,
+      "loss": 1.3409,
+      "step": 534
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019966934353573258,
+      "loss": 1.2021,
+      "step": 535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001996665250191443,
+      "loss": 1.6676,
+      "step": 536
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019966369456102302,
+      "loss": 1.6856,
+      "step": 537
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019966085216170787,
+      "loss": 1.4836,
+      "step": 538
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019965799782153945,
+      "loss": 1.3184,
+      "step": 539
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001996551315408597,
+      "loss": 1.5461,
+      "step": 540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019965225332001205,
+      "loss": 1.1585,
+      "step": 541
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019964936315934137,
+      "loss": 1.2554,
+      "step": 542
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019964646105919394,
+      "loss": 1.0617,
+      "step": 543
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019964354701991748,
+      "loss": 1.4379,
+      "step": 544
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019964062104186115,
+      "loss": 1.4927,
+      "step": 545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019963768312537548,
+      "loss": 1.4482,
+      "step": 546
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019963473327081247,
+      "loss": 1.2271,
+      "step": 547
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019963177147852565,
+      "loss": 1.507,
+      "step": 548
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001996287977488698,
+      "loss": 1.2538,
+      "step": 549
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019962581208220125,
+      "loss": 1.1785,
+      "step": 550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001996228144788777,
+      "loss": 1.105,
+      "step": 551
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019961980493925834,
+      "loss": 1.0748,
+      "step": 552
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019961678346370375,
+      "loss": 1.522,
+      "step": 553
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019961375005257592,
+      "loss": 1.4313,
+      "step": 554
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019961070470623832,
+      "loss": 1.172,
+      "step": 555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019960764742505584,
+      "loss": 1.6879,
+      "step": 556
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001996045782093948,
+      "loss": 1.4415,
+      "step": 557
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001996014970596229,
+      "loss": 1.1283,
+      "step": 558
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001995984039761093,
+      "loss": 1.7704,
+      "step": 559
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019959529895922465,
+      "loss": 1.6023,
+      "step": 560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019959218200934088,
+      "loss": 1.5484,
+      "step": 561
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001995890531268316,
+      "loss": 1.8002,
+      "step": 562
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019958591231207154,
+      "loss": 1.3725,
+      "step": 563
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001995827595654371,
+      "loss": 1.3675,
+      "step": 564
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019957959488730605,
+      "loss": 1.5015,
+      "step": 565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019957641827805748,
+      "loss": 1.2965,
+      "step": 566
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019957322973807208,
+      "loss": 1.4661,
+      "step": 567
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019957002926773184,
+      "loss": 1.2713,
+      "step": 568
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001995668168674202,
+      "loss": 1.3112,
+      "step": 569
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001995635925375221,
+      "loss": 1.0394,
+      "step": 570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019956035627842383,
+      "loss": 1.6115,
+      "step": 571
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001995571080905132,
+      "loss": 1.2549,
+      "step": 572
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019955384797417934,
+      "loss": 1.2809,
+      "step": 573
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019955057592981285,
+      "loss": 1.2021,
+      "step": 574
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001995472919578058,
+      "loss": 1.2858,
+      "step": 575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019954399605855166,
+      "loss": 1.632,
+      "step": 576
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019954068823244532,
+      "loss": 1.446,
+      "step": 577
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019953736847988306,
+      "loss": 1.2286,
+      "step": 578
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001995340368012627,
+      "loss": 1.4105,
+      "step": 579
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019953069319698344,
+      "loss": 1.3658,
+      "step": 580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001995273376674458,
+      "loss": 1.0483,
+      "step": 581
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019952397021305194,
+      "loss": 1.2664,
+      "step": 582
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019952059083420522,
+      "loss": 1.6795,
+      "step": 583
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019951719953131063,
+      "loss": 1.2621,
+      "step": 584
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019951379630477444,
+      "loss": 1.634,
+      "step": 585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001995103811550044,
+      "loss": 1.2667,
+      "step": 586
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019950695408240977,
+      "loss": 1.2274,
+      "step": 587
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019950351508740108,
+      "loss": 1.4132,
+      "step": 588
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019950006417039043,
+      "loss": 1.5447,
+      "step": 589
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019949660133179126,
+      "loss": 1.4938,
+      "step": 590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001994931265720185,
+      "loss": 1.4523,
+      "step": 591
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019948963989148844,
+      "loss": 1.7052,
+      "step": 592
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019948614129061887,
+      "loss": 1.3599,
+      "step": 593
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019948263076982893,
+      "loss": 1.5975,
+      "step": 594
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019947910832953926,
+      "loss": 1.3507,
+      "step": 595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019947557397017194,
+      "loss": 1.332,
+      "step": 596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019947202769215038,
+      "loss": 1.2992,
+      "step": 597
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019946846949589947,
+      "loss": 1.4824,
+      "step": 598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019946489938184564,
+      "loss": 1.3501,
+      "step": 599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001994613173504165,
+      "loss": 1.1946,
+      "step": 600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019945772340204128,
+      "loss": 1.3337,
+      "step": 601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019945411753715067,
+      "loss": 1.5463,
+      "step": 602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001994504997561766,
+      "loss": 1.6296,
+      "step": 603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001994468700595526,
+      "loss": 1.3266,
+      "step": 604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001994432284477135,
+      "loss": 1.2063,
+      "step": 605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001994395749210957,
+      "loss": 0.9311,
+      "step": 606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019943590948013685,
+      "loss": 1.5757,
+      "step": 607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019943223212527623,
+      "loss": 1.1268,
+      "step": 608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019942854285695435,
+      "loss": 1.461,
+      "step": 609
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019942484167561331,
+      "loss": 1.4543,
+      "step": 610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019942112858169653,
+      "loss": 1.4428,
+      "step": 611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019941740357564886,
+      "loss": 1.494,
+      "step": 612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019941366665791668,
+      "loss": 1.3247,
+      "step": 613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019940991782894772,
+      "loss": 1.4623,
+      "step": 614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001994061570891911,
+      "loss": 1.4784,
+      "step": 615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019940238443909742,
+      "loss": 1.3462,
+      "step": 616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019939859987911875,
+      "loss": 1.6669,
+      "step": 617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019939480340970848,
+      "loss": 1.3546,
+      "step": 618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001993909950313215,
+      "loss": 1.1084,
+      "step": 619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019938717474441412,
+      "loss": 1.3459,
+      "step": 620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019938334254944408,
+      "loss": 1.0903,
+      "step": 621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019937949844687052,
+      "loss": 1.3014,
+      "step": 622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019937564243715402,
+      "loss": 1.3839,
+      "step": 623
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019937177452075658,
+      "loss": 1.5945,
+      "step": 624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001993678946981416,
+      "loss": 1.2057,
+      "step": 625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019936400296977406,
+      "loss": 1.7025,
+      "step": 626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001993600993361201,
+      "loss": 1.4723,
+      "step": 627
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019935618379764752,
+      "loss": 1.3695,
+      "step": 628
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019935225635482544,
+      "loss": 1.231,
+      "step": 629
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019934831700812444,
+      "loss": 1.8425,
+      "step": 630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001993443657580165,
+      "loss": 1.5978,
+      "step": 631
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019934040260497502,
+      "loss": 1.2074,
+      "step": 632
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019933642754947485,
+      "loss": 1.5021,
+      "step": 633
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001993324405919923,
+      "loss": 1.2223,
+      "step": 634
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019932844173300505,
+      "loss": 1.4719,
+      "step": 635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001993244309729922,
+      "loss": 1.4559,
+      "step": 636
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001993204083124343,
+      "loss": 1.3343,
+      "step": 637
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019931637375181333,
+      "loss": 1.3801,
+      "step": 638
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019931232729161272,
+      "loss": 1.6145,
+      "step": 639
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019930826893231726,
+      "loss": 1.3321,
+      "step": 640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001993041986744132,
+      "loss": 1.4667,
+      "step": 641
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019930011651838827,
+      "loss": 1.1638,
+      "step": 642
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001992960224647315,
+      "loss": 1.6957,
+      "step": 643
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019929191651393347,
+      "loss": 1.1194,
+      "step": 644
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019928779866648611,
+      "loss": 1.3774,
+      "step": 645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019928366892288285,
+      "loss": 1.042,
+      "step": 646
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001992795272836184,
+      "loss": 1.4633,
+      "step": 647
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019927537374918902,
+      "loss": 1.228,
+      "step": 648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019927120832009246,
+      "loss": 1.3016,
+      "step": 649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019926703099682768,
+      "loss": 1.2221,
+      "step": 650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019926284177989524,
+      "loss": 1.2888,
+      "step": 651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019925864066979706,
+      "loss": 1.3031,
+      "step": 652
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001992544276670365,
+      "loss": 1.2724,
+      "step": 653
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019925020277211837,
+      "loss": 1.6188,
+      "step": 654
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019924596598554881,
+      "loss": 1.4181,
+      "step": 655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019924171730783548,
+      "loss": 1.3763,
+      "step": 656
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019923745673948746,
+      "loss": 1.4518,
+      "step": 657
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001992331842810152,
+      "loss": 1.3499,
+      "step": 658
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019922889993293064,
+      "loss": 1.2693,
+      "step": 659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019922460369574703,
+      "loss": 1.2688,
+      "step": 660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001992202955699792,
+      "loss": 1.1912,
+      "step": 661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001992159755561433,
+      "loss": 1.1234,
+      "step": 662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019921164365475696,
+      "loss": 1.437,
+      "step": 663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019920729986633914,
+      "loss": 1.3476,
+      "step": 664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019920294419141038,
+      "loss": 1.2374,
+      "step": 665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019919857663049247,
+      "loss": 1.5081,
+      "step": 666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019919419718410875,
+      "loss": 1.3102,
+      "step": 667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019918980585278395,
+      "loss": 1.4894,
+      "step": 668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001991854026370442,
+      "loss": 1.1372,
+      "step": 669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019918098753741705,
+      "loss": 1.6508,
+      "step": 670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019917656055443156,
+      "loss": 1.251,
+      "step": 671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019917212168861807,
+      "loss": 1.4552,
+      "step": 672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019916767094050849,
+      "loss": 1.2616,
+      "step": 673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019916320831063605,
+      "loss": 1.3403,
+      "step": 674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019915873379953544,
+      "loss": 1.0344,
+      "step": 675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019915424740774278,
+      "loss": 1.5404,
+      "step": 676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001991497491357956,
+      "loss": 1.8023,
+      "step": 677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019914523898423287,
+      "loss": 1.2658,
+      "step": 678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019914071695359498,
+      "loss": 1.4508,
+      "step": 679
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001991361830444237,
+      "loss": 1.4968,
+      "step": 680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001991316372572623,
+      "loss": 1.27,
+      "step": 681
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019912707959265542,
+      "loss": 1.3341,
+      "step": 682
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001991225100511491,
+      "loss": 1.1784,
+      "step": 683
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019911792863329094,
+      "loss": 1.1617,
+      "step": 684
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019911333533962977,
+      "loss": 1.4568,
+      "step": 685
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019910873017071594,
+      "loss": 1.3171,
+      "step": 686
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019910411312710124,
+      "loss": 1.3737,
+      "step": 687
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019909948420933886,
+      "loss": 1.0699,
+      "step": 688
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019909484341798343,
+      "loss": 1.4607,
+      "step": 689
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019909019075359098,
+      "loss": 0.9941,
+      "step": 690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019908552621671893,
+      "loss": 1.3217,
+      "step": 691
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019908084980792618,
+      "loss": 1.5066,
+      "step": 692
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019907616152777303,
+      "loss": 0.9421,
+      "step": 693
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019907146137682123,
+      "loss": 1.3429,
+      "step": 694
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001990667493556339,
+      "loss": 1.5841,
+      "step": 695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019906202546477565,
+      "loss": 1.4733,
+      "step": 696
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019905728970481246,
+      "loss": 1.5315,
+      "step": 697
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019905254207631173,
+      "loss": 1.512,
+      "step": 698
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019904778257984226,
+      "loss": 1.5503,
+      "step": 699
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019904301121597437,
+      "loss": 1.5544,
+      "step": 700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001990382279852797,
+      "loss": 1.4181,
+      "step": 701
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019903343288833137,
+      "loss": 1.1761,
+      "step": 702
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019902862592570392,
+      "loss": 1.4651,
+      "step": 703
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019902380709797325,
+      "loss": 0.952,
+      "step": 704
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019901897640571678,
+      "loss": 1.3676,
+      "step": 705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019901413384951328,
+      "loss": 1.2363,
+      "step": 706
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019900927942994296,
+      "loss": 1.5995,
+      "step": 707
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019900441314758745,
+      "loss": 1.6623,
+      "step": 708
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019899953500302978,
+      "loss": 1.569,
+      "step": 709
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019899464499685443,
+      "loss": 1.2317,
+      "step": 710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019898974312964734,
+      "loss": 1.2427,
+      "step": 711
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019898482940199576,
+      "loss": 1.2767,
+      "step": 712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019897990381448848,
+      "loss": 1.2083,
+      "step": 713
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019897496636771568,
+      "loss": 1.2245,
+      "step": 714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019897001706226886,
+      "loss": 1.1779,
+      "step": 715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019896505589874107,
+      "loss": 1.5035,
+      "step": 716
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019896008287772671,
+      "loss": 1.3512,
+      "step": 717
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019895509799982167,
+      "loss": 1.3017,
+      "step": 718
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019895010126562315,
+      "loss": 1.3244,
+      "step": 719
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019894509267572987,
+      "loss": 1.0487,
+      "step": 720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019894007223074194,
+      "loss": 1.4245,
+      "step": 721
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019893503993126085,
+      "loss": 1.5963,
+      "step": 722
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019892999577788957,
+      "loss": 1.2694,
+      "step": 723
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019892493977123247,
+      "loss": 1.2235,
+      "step": 724
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001989198719118953,
+      "loss": 1.5386,
+      "step": 725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001989147922004853,
+      "loss": 0.9936,
+      "step": 726
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001989097006376111,
+      "loss": 1.292,
+      "step": 727
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001989045972238827,
+      "loss": 1.3865,
+      "step": 728
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019889948195991162,
+      "loss": 1.709,
+      "step": 729
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001988943548463107,
+      "loss": 1.6039,
+      "step": 730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019888921588369425,
+      "loss": 1.472,
+      "step": 731
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019888406507267802,
+      "loss": 1.4074,
+      "step": 732
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019887890241387915,
+      "loss": 1.556,
+      "step": 733
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019887372790791617,
+      "loss": 1.5555,
+      "step": 734
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001988685415554091,
+      "loss": 1.4138,
+      "step": 735
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019886334335697932,
+      "loss": 1.1593,
+      "step": 736
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019885813331324968,
+      "loss": 1.3493,
+      "step": 737
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019885291142484438,
+      "loss": 1.2099,
+      "step": 738
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001988476776923891,
+      "loss": 1.2362,
+      "step": 739
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019884243211651094,
+      "loss": 1.174,
+      "step": 740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019883717469783836,
+      "loss": 1.3776,
+      "step": 741
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019883190543700132,
+      "loss": 1.3016,
+      "step": 742
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001988266243346311,
+      "loss": 1.4384,
+      "step": 743
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001988213313913605,
+      "loss": 1.293,
+      "step": 744
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019881602660782368,
+      "loss": 1.2847,
+      "step": 745
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019881070998465622,
+      "loss": 1.211,
+      "step": 746
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019880538152249516,
+      "loss": 1.0661,
+      "step": 747
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001988000412219789,
+      "loss": 1.2637,
+      "step": 748
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001987946890837473,
+      "loss": 1.2507,
+      "step": 749
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001987893251084416,
+      "loss": 0.976,
+      "step": 750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019878394929670457,
+      "loss": 1.3686,
+      "step": 751
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001987785616491802,
+      "loss": 1.2519,
+      "step": 752
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019877316216651404,
+      "loss": 1.4407,
+      "step": 753
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001987677508493531,
+      "loss": 1.178,
+      "step": 754
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019876232769834568,
+      "loss": 1.2569,
+      "step": 755
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019875689271414153,
+      "loss": 1.1282,
+      "step": 756
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001987514458973919,
+      "loss": 1.7222,
+      "step": 757
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019874598724874934,
+      "loss": 1.1672,
+      "step": 758
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019874051676886794,
+      "loss": 1.5353,
+      "step": 759
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019873503445840308,
+      "loss": 1.3421,
+      "step": 760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001987295403180117,
+      "loss": 1.3226,
+      "step": 761
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000198724034348352,
+      "loss": 1.4656,
+      "step": 762
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001987185165500837,
+      "loss": 0.9286,
+      "step": 763
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019871298692386796,
+      "loss": 1.31,
+      "step": 764
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019870744547036728,
+      "loss": 1.5787,
+      "step": 765
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019870189219024559,
+      "loss": 1.1258,
+      "step": 766
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019869632708416829,
+      "loss": 1.4453,
+      "step": 767
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001986907501528021,
+      "loss": 1.6193,
+      "step": 768
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001986851613968153,
+      "loss": 1.3045,
+      "step": 769
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019867956081687747,
+      "loss": 1.355,
+      "step": 770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019867394841365966,
+      "loss": 1.3986,
+      "step": 771
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019866832418783428,
+      "loss": 1.52,
+      "step": 772
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019866268814007523,
+      "loss": 1.2962,
+      "step": 773
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019865704027105776,
+      "loss": 1.3421,
+      "step": 774
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001986513805814586,
+      "loss": 1.2923,
+      "step": 775
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001986457090719559,
+      "loss": 1.5693,
+      "step": 776
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001986400257432291,
+      "loss": 1.5755,
+      "step": 777
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019863433059595916,
+      "loss": 1.3285,
+      "step": 778
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019862862363082852,
+      "loss": 1.2638,
+      "step": 779
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001986229048485209,
+      "loss": 1.1303,
+      "step": 780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019861717424972156,
+      "loss": 1.2136,
+      "step": 781
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019861143183511702,
+      "loss": 1.7529,
+      "step": 782
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019860567760539538,
+      "loss": 1.4082,
+      "step": 783
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019859991156124602,
+      "loss": 1.0247,
+      "step": 784
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019859413370335982,
+      "loss": 1.6251,
+      "step": 785
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019858834403242907,
+      "loss": 1.4077,
+      "step": 786
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019858254254914747,
+      "loss": 1.329,
+      "step": 787
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001985767292542101,
+      "loss": 1.3361,
+      "step": 788
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001985709041483135,
+      "loss": 1.4612,
+      "step": 789
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019856506723215557,
+      "loss": 1.5562,
+      "step": 790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001985592185064357,
+      "loss": 1.3301,
+      "step": 791
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001985533579718546,
+      "loss": 1.4059,
+      "step": 792
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001985474856291145,
+      "loss": 1.2752,
+      "step": 793
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000198541601478919,
+      "loss": 1.3308,
+      "step": 794
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019853570552197305,
+      "loss": 1.4371,
+      "step": 795
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001985297977589831,
+      "loss": 0.8591,
+      "step": 796
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019852387819065703,
+      "loss": 1.4759,
+      "step": 797
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019851794681770404,
+      "loss": 1.3288,
+      "step": 798
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001985120036408348,
+      "loss": 1.2994,
+      "step": 799
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019850604866076147,
+      "loss": 1.6827,
+      "step": 800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019850008187819745,
+      "loss": 1.4214,
+      "step": 801
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019849410329385764,
+      "loss": 1.2451,
+      "step": 802
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019848811290845846,
+      "loss": 1.4478,
+      "step": 803
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019848211072271758,
+      "loss": 1.0902,
+      "step": 804
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019847609673735413,
+      "loss": 1.2861,
+      "step": 805
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019847007095308877,
+      "loss": 1.0253,
+      "step": 806
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019846403337064336,
+      "loss": 1.3274,
+      "step": 807
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019845798399074135,
+      "loss": 1.5834,
+      "step": 808
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001984519228141076,
+      "loss": 1.1784,
+      "step": 809
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019844584984146822,
+      "loss": 1.4456,
+      "step": 810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019843976507355091,
+      "loss": 1.2794,
+      "step": 811
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001984336685110847,
+      "loss": 1.0854,
+      "step": 812
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019842756015480006,
+      "loss": 1.1527,
+      "step": 813
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019842144000542885,
+      "loss": 1.1002,
+      "step": 814
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019841530806370436,
+      "loss": 1.2387,
+      "step": 815
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001984091643303613,
+      "loss": 1.7089,
+      "step": 816
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019840300880613577,
+      "loss": 1.4764,
+      "step": 817
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019839684149176526,
+      "loss": 1.4814,
+      "step": 818
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019839066238798875,
+      "loss": 1.092,
+      "step": 819
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019838447149554656,
+      "loss": 1.2631,
+      "step": 820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001983782688151805,
+      "loss": 1.2801,
+      "step": 821
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001983720543476337,
+      "loss": 1.7286,
+      "step": 822
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019836582809365071,
+      "loss": 1.3466,
+      "step": 823
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001983595900539776,
+      "loss": 1.3982,
+      "step": 824
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001983533402293618,
+      "loss": 1.3596,
+      "step": 825
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019834707862055205,
+      "loss": 1.1853,
+      "step": 826
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001983408052282986,
+      "loss": 1.8524,
+      "step": 827
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001983345200533531,
+      "loss": 1.4887,
+      "step": 828
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019832822309646866,
+      "loss": 1.4422,
+      "step": 829
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019832191435839973,
+      "loss": 1.5869,
+      "step": 830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019831559383990213,
+      "loss": 1.288,
+      "step": 831
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001983092615417332,
+      "loss": 1.1484,
+      "step": 832
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019830291746465167,
+      "loss": 1.5971,
+      "step": 833
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001982965616094176,
+      "loss": 1.0338,
+      "step": 834
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019829019397679253,
+      "loss": 1.275,
+      "step": 835
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019828381456753942,
+      "loss": 1.3898,
+      "step": 836
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019827742338242259,
+      "loss": 1.473,
+      "step": 837
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001982710204222078,
+      "loss": 1.361,
+      "step": 838
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019826460568766224,
+      "loss": 0.9665,
+      "step": 839
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019825817917955453,
+      "loss": 1.24,
+      "step": 840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019825174089865456,
+      "loss": 1.1623,
+      "step": 841
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019824529084573383,
+      "loss": 1.244,
+      "step": 842
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001982388290215651,
+      "loss": 1.5797,
+      "step": 843
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019823235542692258,
+      "loss": 1.3055,
+      "step": 844
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019822587006258194,
+      "loss": 1.482,
+      "step": 845
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001982193729293202,
+      "loss": 1.2161,
+      "step": 846
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001982128640279158,
+      "loss": 1.0643,
+      "step": 847
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019820634335914868,
+      "loss": 1.2399,
+      "step": 848
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001981998109238,
+      "loss": 1.4426,
+      "step": 849
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019819326672265255,
+      "loss": 1.2758,
+      "step": 850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019818671075649035,
+      "loss": 1.4923,
+      "step": 851
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019818014302609892,
+      "loss": 1.2974,
+      "step": 852
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001981735635322652,
+      "loss": 1.2119,
+      "step": 853
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019816697227577747,
+      "loss": 1.0295,
+      "step": 854
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019816036925742546,
+      "loss": 1.3857,
+      "step": 855
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019815375447800038,
+      "loss": 1.2346,
+      "step": 856
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001981471279382947,
+      "loss": 1.4461,
+      "step": 857
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001981404896391024,
+      "loss": 1.2407,
+      "step": 858
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019813383958121886,
+      "loss": 1.2795,
+      "step": 859
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019812717776544086,
+      "loss": 1.5569,
+      "step": 860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019812050419256653,
+      "loss": 1.7648,
+      "step": 861
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019811381886339555,
+      "loss": 1.2435,
+      "step": 862
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019810712177872888,
+      "loss": 1.6191,
+      "step": 863
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019810041293936891,
+      "loss": 1.108,
+      "step": 864
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001980936923461195,
+      "loss": 1.3631,
+      "step": 865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019808695999978584,
+      "loss": 1.204,
+      "step": 866
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019808021590117457,
+      "loss": 1.7135,
+      "step": 867
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019807346005109376,
+      "loss": 1.2119,
+      "step": 868
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019806669245035286,
+      "loss": 1.583,
+      "step": 869
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019805991309976268,
+      "loss": 1.3601,
+      "step": 870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019805312200013556,
+      "loss": 1.3342,
+      "step": 871
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019804631915228512,
+      "loss": 1.6129,
+      "step": 872
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019803950455702646,
+      "loss": 1.6823,
+      "step": 873
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019803267821517605,
+      "loss": 1.0572,
+      "step": 874
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019802584012755184,
+      "loss": 1.3556,
+      "step": 875
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001980189902949731,
+      "loss": 1.4452,
+      "step": 876
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019801212871826048,
+      "loss": 1.5821,
+      "step": 877
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019800525539823624,
+      "loss": 1.099,
+      "step": 878
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001979983703357238,
+      "loss": 1.7666,
+      "step": 879
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019799147353154813,
+      "loss": 1.0839,
+      "step": 880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019798456498653554,
+      "loss": 1.3458,
+      "step": 881
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001979776447015138,
+      "loss": 1.7073,
+      "step": 882
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019797071267731208,
+      "loss": 1.5211,
+      "step": 883
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001979637689147609,
+      "loss": 1.3359,
+      "step": 884
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019795681341469226,
+      "loss": 1.7509,
+      "step": 885
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019794984617793954,
+      "loss": 1.3011,
+      "step": 886
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019794286720533743,
+      "loss": 1.3744,
+      "step": 887
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019793587649772224,
+      "loss": 1.4663,
+      "step": 888
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019792887405593147,
+      "loss": 1.3806,
+      "step": 889
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001979218598808042,
+      "loss": 1.4385,
+      "step": 890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001979148339731807,
+      "loss": 1.4471,
+      "step": 891
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019790779633390294,
+      "loss": 1.1925,
+      "step": 892
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019790074696381403,
+      "loss": 1.3289,
+      "step": 893
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001978936858637586,
+      "loss": 1.6461,
+      "step": 894
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001978866130345827,
+      "loss": 1.3027,
+      "step": 895
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019787952847713373,
+      "loss": 1.6242,
+      "step": 896
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019787243219226058,
+      "loss": 1.2409,
+      "step": 897
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001978653241808134,
+      "loss": 1.3378,
+      "step": 898
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019785820444364393,
+      "loss": 1.4628,
+      "step": 899
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001978510729816052,
+      "loss": 1.116,
+      "step": 900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019784392979555162,
+      "loss": 1.1478,
+      "step": 901
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019783677488633908,
+      "loss": 1.5398,
+      "step": 902
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019782960825482482,
+      "loss": 1.3803,
+      "step": 903
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019782242990186755,
+      "loss": 1.5924,
+      "step": 904
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019781523982832734,
+      "loss": 1.1896,
+      "step": 905
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019780803803506563,
+      "loss": 1.4936,
+      "step": 906
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019780082452294532,
+      "loss": 1.1102,
+      "step": 907
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019779359929283073,
+      "loss": 1.2779,
+      "step": 908
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019778636234558746,
+      "loss": 1.593,
+      "step": 909
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019777911368208271,
+      "loss": 1.2976,
+      "step": 910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019777185330318493,
+      "loss": 1.1286,
+      "step": 911
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019776458120976403,
+      "loss": 1.4294,
+      "step": 912
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001977572974026913,
+      "loss": 1.3558,
+      "step": 913
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019775000188283943,
+      "loss": 1.1965,
+      "step": 914
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019774269465108258,
+      "loss": 1.5507,
+      "step": 915
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019773537570829622,
+      "loss": 1.1036,
+      "step": 916
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001977280450553573,
+      "loss": 1.1269,
+      "step": 917
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019772070269314415,
+      "loss": 1.499,
+      "step": 918
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001977133486225365,
+      "loss": 1.2427,
+      "step": 919
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019770598284441542,
+      "loss": 1.5038,
+      "step": 920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001976986053596635,
+      "loss": 1.1728,
+      "step": 921
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001976912161691646,
+      "loss": 1.7111,
+      "step": 922
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019768381527380413,
+      "loss": 1.2663,
+      "step": 923
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001976764026744688,
+      "loss": 1.6135,
+      "step": 924
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019766897837204675,
+      "loss": 1.1838,
+      "step": 925
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019766154236742752,
+      "loss": 1.225,
+      "step": 926
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019765409466150204,
+      "loss": 1.4209,
+      "step": 927
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019764663525516268,
+      "loss": 1.4197,
+      "step": 928
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019763916414930317,
+      "loss": 1.5356,
+      "step": 929
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019763168134481868,
+      "loss": 1.1513,
+      "step": 930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019762418684260576,
+      "loss": 1.3176,
+      "step": 931
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019761668064356233,
+      "loss": 1.3302,
+      "step": 932
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019760916274858778,
+      "loss": 1.1717,
+      "step": 933
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019760163315858288,
+      "loss": 1.4058,
+      "step": 934
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019759409187444972,
+      "loss": 1.1105,
+      "step": 935
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019758653889709188,
+      "loss": 1.1785,
+      "step": 936
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019757897422741438,
+      "loss": 1.4589,
+      "step": 937
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001975713978663235,
+      "loss": 1.1989,
+      "step": 938
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019756380981472707,
+      "loss": 1.2819,
+      "step": 939
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001975562100735342,
+      "loss": 1.2346,
+      "step": 940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019754859864365548,
+      "loss": 1.5419,
+      "step": 941
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019754097552600287,
+      "loss": 1.3385,
+      "step": 942
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001975333407214897,
+      "loss": 1.2023,
+      "step": 943
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019752569423103078,
+      "loss": 1.1564,
+      "step": 944
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019751803605554225,
+      "loss": 1.5928,
+      "step": 945
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019751036619594168,
+      "loss": 1.0149,
+      "step": 946
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019750268465314804,
+      "loss": 1.3728,
+      "step": 947
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019749499142808166,
+      "loss": 1.6249,
+      "step": 948
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019748728652166432,
+      "loss": 1.2277,
+      "step": 949
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019747956993481923,
+      "loss": 1.2682,
+      "step": 950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019747184166847085,
+      "loss": 1.2162,
+      "step": 951
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019746410172354523,
+      "loss": 1.2932,
+      "step": 952
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019745635010096971,
+      "loss": 1.1025,
+      "step": 953
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019744858680167306,
+      "loss": 1.3241,
+      "step": 954
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001974408118265854,
+      "loss": 1.2649,
+      "step": 955
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001974330251766383,
+      "loss": 1.5343,
+      "step": 956
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019742522685276472,
+      "loss": 1.3406,
+      "step": 957
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019741741685589905,
+      "loss": 1.6525,
+      "step": 958
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000197409595186977,
+      "loss": 1.5102,
+      "step": 959
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019740176184693573,
+      "loss": 1.5713,
+      "step": 960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001973939168367138,
+      "loss": 0.9938,
+      "step": 961
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019738606015725117,
+      "loss": 1.479,
+      "step": 962
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019737819180948914,
+      "loss": 1.0533,
+      "step": 963
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001973703117943705,
+      "loss": 1.309,
+      "step": 964
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001973624201128394,
+      "loss": 1.0782,
+      "step": 965
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019735451676584136,
+      "loss": 1.2671,
+      "step": 966
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001973466017543233,
+      "loss": 1.3563,
+      "step": 967
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019733867507923362,
+      "loss": 1.6589,
+      "step": 968
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019733073674152197,
+      "loss": 1.5296,
+      "step": 969
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019732278674213953,
+      "loss": 1.1417,
+      "step": 970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019731482508203883,
+      "loss": 1.3193,
+      "step": 971
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019730685176217377,
+      "loss": 1.2853,
+      "step": 972
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019729886678349968,
+      "loss": 1.4187,
+      "step": 973
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001972908701469733,
+      "loss": 1.6301,
+      "step": 974
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019728286185355276,
+      "loss": 1.4625,
+      "step": 975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001972748419041975,
+      "loss": 1.7036,
+      "step": 976
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019726681029986852,
+      "loss": 1.3678,
+      "step": 977
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019725876704152805,
+      "loss": 1.2169,
+      "step": 978
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019725071213013985,
+      "loss": 1.4715,
+      "step": 979
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019724264556666895,
+      "loss": 1.4902,
+      "step": 980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019723456735208192,
+      "loss": 1.1793,
+      "step": 981
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001972264774873466,
+      "loss": 1.74,
+      "step": 982
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001972183759734323,
+      "loss": 1.0887,
+      "step": 983
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001972102628113097,
+      "loss": 1.0665,
+      "step": 984
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019720213800195088,
+      "loss": 1.4826,
+      "step": 985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019719400154632926,
+      "loss": 1.1575,
+      "step": 986
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001971858534454198,
+      "loss": 1.0979,
+      "step": 987
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001971776937001987,
+      "loss": 1.4039,
+      "step": 988
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001971695223116436,
+      "loss": 1.2287,
+      "step": 989
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019716133928073364,
+      "loss": 1.4117,
+      "step": 990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001971531446084492,
+      "loss": 1.2672,
+      "step": 991
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019714493829577214,
+      "loss": 1.5899,
+      "step": 992
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001971367203436857,
+      "loss": 1.5006,
+      "step": 993
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019712849075317454,
+      "loss": 1.7531,
+      "step": 994
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019712024952522463,
+      "loss": 1.3746,
+      "step": 995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019711199666082343,
+      "loss": 1.3079,
+      "step": 996
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019710373216095978,
+      "loss": 1.3198,
+      "step": 997
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019709545602662382,
+      "loss": 1.1947,
+      "step": 998
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001970871682588072,
+      "loss": 1.5583,
+      "step": 999
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001970788688585029,
+      "loss": 0.9531,
+      "step": 1000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019707055782670535,
+      "loss": 1.4942,
+      "step": 1001
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001970622351644103,
+      "loss": 1.4301,
+      "step": 1002
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019705390087261494,
+      "loss": 1.1768,
+      "step": 1003
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019704555495231782,
+      "loss": 1.5109,
+      "step": 1004
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019703719740451892,
+      "loss": 1.0715,
+      "step": 1005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019702882823021965,
+      "loss": 1.31,
+      "step": 1006
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019702044743042266,
+      "loss": 1.2104,
+      "step": 1007
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019701205500613218,
+      "loss": 1.3009,
+      "step": 1008
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019700365095835372,
+      "loss": 1.2368,
+      "step": 1009
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019699523528809418,
+      "loss": 1.0904,
+      "step": 1010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019698680799636194,
+      "loss": 1.1986,
+      "step": 1011
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019697836908416668,
+      "loss": 1.2936,
+      "step": 1012
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019696991855251946,
+      "loss": 1.59,
+      "step": 1013
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001969614564024329,
+      "loss": 1.1161,
+      "step": 1014
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001969529826349208,
+      "loss": 1.8513,
+      "step": 1015
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019694449725099844,
+      "loss": 1.4476,
+      "step": 1016
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019693600025168256,
+      "loss": 1.5224,
+      "step": 1017
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019692749163799115,
+      "loss": 1.6785,
+      "step": 1018
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019691897141094374,
+      "loss": 1.4652,
+      "step": 1019
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019691043957156113,
+      "loss": 1.2173,
+      "step": 1020
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019690189612086556,
+      "loss": 1.2681,
+      "step": 1021
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001968933410598807,
+      "loss": 1.6668,
+      "step": 1022
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019688477438963156,
+      "loss": 1.3419,
+      "step": 1023
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019687619611114454,
+      "loss": 1.1165,
+      "step": 1024
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019686760622544746,
+      "loss": 0.9762,
+      "step": 1025
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019685900473356951,
+      "loss": 1.5937,
+      "step": 1026
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019685039163654127,
+      "loss": 1.1013,
+      "step": 1027
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019684176693539473,
+      "loss": 1.2142,
+      "step": 1028
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019683313063116327,
+      "loss": 1.516,
+      "step": 1029
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019682448272488162,
+      "loss": 1.3197,
+      "step": 1030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001968158232175859,
+      "loss": 1.2921,
+      "step": 1031
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019680715211031373,
+      "loss": 1.0612,
+      "step": 1032
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019679846940410395,
+      "loss": 1.2144,
+      "step": 1033
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019678977509999697,
+      "loss": 1.5292,
+      "step": 1034
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001967810691990344,
+      "loss": 1.475,
+      "step": 1035
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019677235170225942,
+      "loss": 1.4802,
+      "step": 1036
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019676362261071646,
+      "loss": 1.3306,
+      "step": 1037
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019675488192545144,
+      "loss": 1.3979,
+      "step": 1038
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001967461296475116,
+      "loss": 1.3994,
+      "step": 1039
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019673736577794553,
+      "loss": 1.1436,
+      "step": 1040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019672859031780337,
+      "loss": 1.5083,
+      "step": 1041
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019671980326813656,
+      "loss": 1.3665,
+      "step": 1042
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001967110046299978,
+      "loss": 1.6811,
+      "step": 1043
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019670219440444143,
+      "loss": 1.3779,
+      "step": 1044
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019669337259252295,
+      "loss": 1.5686,
+      "step": 1045
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019668453919529941,
+      "loss": 1.2419,
+      "step": 1046
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019667569421382913,
+      "loss": 1.6625,
+      "step": 1047
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001966668376491719,
+      "loss": 1.3276,
+      "step": 1048
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019665796950238892,
+      "loss": 1.1714,
+      "step": 1049
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019664908977454258,
+      "loss": 1.5072,
+      "step": 1050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019664019846669698,
+      "loss": 1.1864,
+      "step": 1051
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019663129557991727,
+      "loss": 0.964,
+      "step": 1052
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019662238111527026,
+      "loss": 1.4125,
+      "step": 1053
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019661345507382397,
+      "loss": 1.3858,
+      "step": 1054
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019660451745664794,
+      "loss": 1.68,
+      "step": 1055
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019659556826481298,
+      "loss": 1.3339,
+      "step": 1056
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019658660749939132,
+      "loss": 1.2247,
+      "step": 1057
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019657763516145663,
+      "loss": 1.5491,
+      "step": 1058
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019656865125208394,
+      "loss": 1.3327,
+      "step": 1059
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019655965577234963,
+      "loss": 2.0438,
+      "step": 1060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001965506487233315,
+      "loss": 1.3909,
+      "step": 1061
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001965416301061087,
+      "loss": 1.4087,
+      "step": 1062
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019653259992176184,
+      "loss": 1.3342,
+      "step": 1063
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019652355817137285,
+      "loss": 1.0653,
+      "step": 1064
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019651450485602508,
+      "loss": 1.1908,
+      "step": 1065
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001965054399768032,
+      "loss": 1.3885,
+      "step": 1066
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019649636353479338,
+      "loss": 1.5274,
+      "step": 1067
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001964872755310831,
+      "loss": 1.5792,
+      "step": 1068
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019647817596676122,
+      "loss": 1.3746,
+      "step": 1069
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019646906484291801,
+      "loss": 1.5202,
+      "step": 1070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001964599421606451,
+      "loss": 1.397,
+      "step": 1071
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019645080792103555,
+      "loss": 0.9174,
+      "step": 1072
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001964416621251838,
+      "loss": 1.0759,
+      "step": 1073
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001964325047741856,
+      "loss": 1.5661,
+      "step": 1074
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019642333586913818,
+      "loss": 1.4346,
+      "step": 1075
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019641415541114005,
+      "loss": 1.3165,
+      "step": 1076
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019640496340129126,
+      "loss": 1.3437,
+      "step": 1077
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019639575984069306,
+      "loss": 1.3549,
+      "step": 1078
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019638654473044822,
+      "loss": 1.2936,
+      "step": 1079
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019637731807166083,
+      "loss": 1.4837,
+      "step": 1080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019636807986543638,
+      "loss": 1.7543,
+      "step": 1081
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019635883011288177,
+      "loss": 1.2236,
+      "step": 1082
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019634956881510522,
+      "loss": 0.9227,
+      "step": 1083
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019634029597321643,
+      "loss": 1.1306,
+      "step": 1084
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019633101158832633,
+      "loss": 1.273,
+      "step": 1085
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019632171566154745,
+      "loss": 1.2128,
+      "step": 1086
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019631240819399344,
+      "loss": 1.0955,
+      "step": 1087
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019630308918677962,
+      "loss": 1.3588,
+      "step": 1088
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019629375864102245,
+      "loss": 1.6131,
+      "step": 1089
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019628441655783986,
+      "loss": 1.6081,
+      "step": 1090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019627506293835124,
+      "loss": 1.559,
+      "step": 1091
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019626569778367724,
+      "loss": 1.2854,
+      "step": 1092
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019625632109493996,
+      "loss": 1.0795,
+      "step": 1093
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019624693287326284,
+      "loss": 1.2281,
+      "step": 1094
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019623753311977078,
+      "loss": 1.5473,
+      "step": 1095
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019622812183559002,
+      "loss": 1.3938,
+      "step": 1096
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019621869902184807,
+      "loss": 1.2692,
+      "step": 1097
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019620926467967404,
+      "loss": 1.4714,
+      "step": 1098
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019619981881019823,
+      "loss": 1.1969,
+      "step": 1099
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019619036141455242,
+      "loss": 1.2226,
+      "step": 1100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019618089249386978,
+      "loss": 1.3178,
+      "step": 1101
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019617141204928475,
+      "loss": 1.375,
+      "step": 1102
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001961619200819333,
+      "loss": 1.2868,
+      "step": 1103
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019615241659295268,
+      "loss": 1.2336,
+      "step": 1104
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019614290158348154,
+      "loss": 1.0604,
+      "step": 1105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019613337505465994,
+      "loss": 1.1932,
+      "step": 1106
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001961238370076293,
+      "loss": 1.311,
+      "step": 1107
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001961142874435324,
+      "loss": 1.3411,
+      "step": 1108
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019610472636351345,
+      "loss": 1.3155,
+      "step": 1109
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000196095153768718,
+      "loss": 1.4349,
+      "step": 1110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019608556966029293,
+      "loss": 1.3718,
+      "step": 1111
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019607597403938665,
+      "loss": 1.4556,
+      "step": 1112
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001960663669071488,
+      "loss": 1.4367,
+      "step": 1113
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019605674826473049,
+      "loss": 1.5102,
+      "step": 1114
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019604711811328413,
+      "loss": 1.2328,
+      "step": 1115
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001960374764539636,
+      "loss": 1.4505,
+      "step": 1116
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019602782328792415,
+      "loss": 1.5374,
+      "step": 1117
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019601815861632229,
+      "loss": 1.5543,
+      "step": 1118
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019600848244031603,
+      "loss": 1.693,
+      "step": 1119
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019599879476106467,
+      "loss": 1.2691,
+      "step": 1120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019598909557972904,
+      "loss": 1.1953,
+      "step": 1121
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019597938489747117,
+      "loss": 1.2468,
+      "step": 1122
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019596966271545456,
+      "loss": 1.2331,
+      "step": 1123
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001959599290348441,
+      "loss": 1.3975,
+      "step": 1124
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000195950183856806,
+      "loss": 1.1316,
+      "step": 1125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019594042718250784,
+      "loss": 1.1985,
+      "step": 1126
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019593065901311868,
+      "loss": 1.6858,
+      "step": 1127
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001959208793498089,
+      "loss": 1.2774,
+      "step": 1128
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001959110881937502,
+      "loss": 1.3099,
+      "step": 1129
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019590128554611573,
+      "loss": 1.341,
+      "step": 1130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019589147140807998,
+      "loss": 1.3862,
+      "step": 1131
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019588164578081884,
+      "loss": 1.2962,
+      "step": 1132
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019587180866550958,
+      "loss": 1.1375,
+      "step": 1133
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001958619600633308,
+      "loss": 1.1354,
+      "step": 1134
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019585209997546255,
+      "loss": 1.5225,
+      "step": 1135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019584222840308616,
+      "loss": 1.3474,
+      "step": 1136
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019583234534738446,
+      "loss": 1.4919,
+      "step": 1137
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019582245080954154,
+      "loss": 1.2331,
+      "step": 1138
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001958125447907429,
+      "loss": 1.2549,
+      "step": 1139
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001958026272921755,
+      "loss": 1.2035,
+      "step": 1140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019579269831502752,
+      "loss": 1.2491,
+      "step": 1141
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019578275786048864,
+      "loss": 1.5832,
+      "step": 1142
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019577280592974984,
+      "loss": 1.4395,
+      "step": 1143
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001957628425240036,
+      "loss": 1.7957,
+      "step": 1144
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019575286764444356,
+      "loss": 1.5337,
+      "step": 1145
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019574288129226496,
+      "loss": 1.3929,
+      "step": 1146
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019573288346866428,
+      "loss": 1.3317,
+      "step": 1147
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001957228741748394,
+      "loss": 1.3492,
+      "step": 1148
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019571285341198958,
+      "loss": 1.435,
+      "step": 1149
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019570282118131546,
+      "loss": 1.5071,
+      "step": 1150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019569277748401906,
+      "loss": 1.2892,
+      "step": 1151
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019568272232130376,
+      "loss": 1.5083,
+      "step": 1152
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001956726556943743,
+      "loss": 1.3991,
+      "step": 1153
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019566257760443684,
+      "loss": 1.0443,
+      "step": 1154
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019565248805269886,
+      "loss": 1.6879,
+      "step": 1155
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019564238704036929,
+      "loss": 1.6909,
+      "step": 1156
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001956322745686583,
+      "loss": 1.5388,
+      "step": 1157
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001956221506387776,
+      "loss": 1.2721,
+      "step": 1158
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001956120152519401,
+      "loss": 1.2572,
+      "step": 1159
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019560186840936026,
+      "loss": 1.2858,
+      "step": 1160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019559171011225375,
+      "loss": 1.4317,
+      "step": 1161
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019558154036183772,
+      "loss": 1.079,
+      "step": 1162
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019557135915933063,
+      "loss": 1.4636,
+      "step": 1163
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019556116650595238,
+      "loss": 1.6825,
+      "step": 1164
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019555096240292418,
+      "loss": 1.8148,
+      "step": 1165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019554074685146862,
+      "loss": 1.4654,
+      "step": 1166
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019553051985280972,
+      "loss": 1.3144,
+      "step": 1167
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019552028140817278,
+      "loss": 1.4737,
+      "step": 1168
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019551003151878455,
+      "loss": 1.3288,
+      "step": 1169
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019549977018587308,
+      "loss": 1.4822,
+      "step": 1170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019548949741066788,
+      "loss": 1.4661,
+      "step": 1171
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019547921319439975,
+      "loss": 1.4321,
+      "step": 1172
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019546891753830086,
+      "loss": 1.2788,
+      "step": 1173
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019545861044360487,
+      "loss": 1.4892,
+      "step": 1174
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019544829191154664,
+      "loss": 1.2927,
+      "step": 1175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019543796194336258,
+      "loss": 1.1044,
+      "step": 1176
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019542762054029024,
+      "loss": 1.5949,
+      "step": 1177
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001954172677035688,
+      "loss": 1.4668,
+      "step": 1178
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019540690343443858,
+      "loss": 1.3735,
+      "step": 1179
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019539652773414146,
+      "loss": 1.1719,
+      "step": 1180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019538614060392054,
+      "loss": 1.6633,
+      "step": 1181
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001953757420450204,
+      "loss": 1.3636,
+      "step": 1182
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019536533205868691,
+      "loss": 1.2815,
+      "step": 1183
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019535491064616737,
+      "loss": 1.8671,
+      "step": 1184
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001953444778087104,
+      "loss": 1.3677,
+      "step": 1185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000195334033547566,
+      "loss": 1.3559,
+      "step": 1186
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019532357786398557,
+      "loss": 1.2382,
+      "step": 1187
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019531311075922183,
+      "loss": 1.306,
+      "step": 1188
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019530263223452893,
+      "loss": 1.379,
+      "step": 1189
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019529214229116235,
+      "loss": 1.7018,
+      "step": 1190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001952816409303789,
+      "loss": 1.3123,
+      "step": 1191
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019527112815343682,
+      "loss": 1.2883,
+      "step": 1192
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019526060396159574,
+      "loss": 1.2261,
+      "step": 1193
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019525006835611653,
+      "loss": 1.1692,
+      "step": 1194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001952395213382616,
+      "loss": 1.1767,
+      "step": 1195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019522896290929457,
+      "loss": 1.2959,
+      "step": 1196
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019521839307048056,
+      "loss": 1.4259,
+      "step": 1197
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019520781182308595,
+      "loss": 1.3963,
+      "step": 1198
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019519721916837852,
+      "loss": 1.342,
+      "step": 1199
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019518661510762746,
+      "loss": 1.3809,
+      "step": 1200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001951759996421033,
+      "loss": 1.3294,
+      "step": 1201
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001951653727730779,
+      "loss": 1.3613,
+      "step": 1202
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019515473450182452,
+      "loss": 1.3059,
+      "step": 1203
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019514408482961782,
+      "loss": 1.2744,
+      "step": 1204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019513342375773377,
+      "loss": 1.4413,
+      "step": 1205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019512275128744967,
+      "loss": 1.1401,
+      "step": 1206
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019511206742004432,
+      "loss": 1.306,
+      "step": 1207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001951013721567978,
+      "loss": 1.5286,
+      "step": 1208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001950906654989915,
+      "loss": 1.0099,
+      "step": 1209
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001950799474479083,
+      "loss": 1.4009,
+      "step": 1210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019506921800483237,
+      "loss": 1.665,
+      "step": 1211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001950584771710492,
+      "loss": 1.1718,
+      "step": 1212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019504772494784576,
+      "loss": 1.225,
+      "step": 1213
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001950369613365103,
+      "loss": 0.9334,
+      "step": 1214
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001950261863383325,
+      "loss": 1.1136,
+      "step": 1215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019501539995460333,
+      "loss": 1.0839,
+      "step": 1216
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019500460218661517,
+      "loss": 1.3283,
+      "step": 1217
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019499379303566174,
+      "loss": 1.803,
+      "step": 1218
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019498297250303815,
+      "loss": 1.2945,
+      "step": 1219
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019497214059004087,
+      "loss": 1.4549,
+      "step": 1220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019496129729796772,
+      "loss": 1.5869,
+      "step": 1221
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019495044262811786,
+      "loss": 1.4042,
+      "step": 1222
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019493957658179184,
+      "loss": 1.4598,
+      "step": 1223
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019492869916029165,
+      "loss": 1.3799,
+      "step": 1224
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019491781036492048,
+      "loss": 1.2273,
+      "step": 1225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000194906910196983,
+      "loss": 1.5095,
+      "step": 1226
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019489599865778523,
+      "loss": 1.4408,
+      "step": 1227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001948850757486345,
+      "loss": 1.2591,
+      "step": 1228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019487414147083956,
+      "loss": 1.6908,
+      "step": 1229
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019486319582571053,
+      "loss": 1.4962,
+      "step": 1230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001948522388145588,
+      "loss": 1.2366,
+      "step": 1231
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019484127043869722,
+      "loss": 1.245,
+      "step": 1232
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019483029069943995,
+      "loss": 1.3178,
+      "step": 1233
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019481929959810253,
+      "loss": 1.3285,
+      "step": 1234
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019480829713600185,
+      "loss": 1.3353,
+      "step": 1235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001947972833144562,
+      "loss": 1.3044,
+      "step": 1236
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019478625813478516,
+      "loss": 1.3459,
+      "step": 1237
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019477522159830972,
+      "loss": 1.2125,
+      "step": 1238
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019476417370635222,
+      "loss": 1.415,
+      "step": 1239
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019475311446023638,
+      "loss": 1.1119,
+      "step": 1240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019474204386128727,
+      "loss": 1.3438,
+      "step": 1241
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001947309619108313,
+      "loss": 1.6039,
+      "step": 1242
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001947198686101962,
+      "loss": 1.3871,
+      "step": 1243
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001947087639607112,
+      "loss": 1.2548,
+      "step": 1244
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019469764796370677,
+      "loss": 1.2473,
+      "step": 1245
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019468652062051476,
+      "loss": 1.1714,
+      "step": 1246
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001946753819324684,
+      "loss": 1.0092,
+      "step": 1247
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019466423190090223,
+      "loss": 1.2717,
+      "step": 1248
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019465307052715224,
+      "loss": 1.4202,
+      "step": 1249
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019464189781255575,
+      "loss": 1.3338,
+      "step": 1250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019463071375845135,
+      "loss": 1.4689,
+      "step": 1251
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001946195183661791,
+      "loss": 1.4021,
+      "step": 1252
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019460831163708035,
+      "loss": 1.3021,
+      "step": 1253
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019459709357249783,
+      "loss": 1.6266,
+      "step": 1254
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019458586417377572,
+      "loss": 1.3871,
+      "step": 1255
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001945746234422593,
+      "loss": 1.4563,
+      "step": 1256
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019456337137929554,
+      "loss": 1.7023,
+      "step": 1257
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019455210798623252,
+      "loss": 1.2652,
+      "step": 1258
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019454083326441978,
+      "loss": 1.0279,
+      "step": 1259
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001945295472152082,
+      "loss": 1.332,
+      "step": 1260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019451824983995005,
+      "loss": 1.315,
+      "step": 1261
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019450694113999883,
+      "loss": 1.2889,
+      "step": 1262
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001944956211167096,
+      "loss": 1.299,
+      "step": 1263
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019448428977143856,
+      "loss": 1.3582,
+      "step": 1264
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001944729471055435,
+      "loss": 1.4775,
+      "step": 1265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001944615931203833,
+      "loss": 1.2862,
+      "step": 1266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019445022781731846,
+      "loss": 1.6252,
+      "step": 1267
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019443885119771066,
+      "loss": 1.5417,
+      "step": 1268
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019442746326292295,
+      "loss": 1.1311,
+      "step": 1269
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019441606401431985,
+      "loss": 1.6334,
+      "step": 1270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001944046534532671,
+      "loss": 1.7092,
+      "step": 1271
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001943932315811319,
+      "loss": 1.2798,
+      "step": 1272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019438179839928273,
+      "loss": 1.2807,
+      "step": 1273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019437035390908945,
+      "loss": 1.2621,
+      "step": 1274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019435889811192332,
+      "loss": 1.3677,
+      "step": 1275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019434743100915685,
+      "loss": 1.5983,
+      "step": 1276
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019433595260216402,
+      "loss": 1.5142,
+      "step": 1277
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001943244628923201,
+      "loss": 1.3634,
+      "step": 1278
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019431296188100175,
+      "loss": 1.2938,
+      "step": 1279
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019430144956958693,
+      "loss": 1.2632,
+      "step": 1280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019428992595945502,
+      "loss": 1.23,
+      "step": 1281
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019427839105198668,
+      "loss": 1.3749,
+      "step": 1282
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019426684484856398,
+      "loss": 1.457,
+      "step": 1283
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019425528735057032,
+      "loss": 1.2478,
+      "step": 1284
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001942437185593905,
+      "loss": 1.5126,
+      "step": 1285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019423213847641056,
+      "loss": 1.4558,
+      "step": 1286
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019422054710301805,
+      "loss": 1.4372,
+      "step": 1287
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001942089444406017,
+      "loss": 1.5689,
+      "step": 1288
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019419733049055177,
+      "loss": 1.3573,
+      "step": 1289
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019418570525425976,
+      "loss": 1.1228,
+      "step": 1290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001941740687331185,
+      "loss": 1.0451,
+      "step": 1291
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019416242092852226,
+      "loss": 1.1049,
+      "step": 1292
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001941507618418666,
+      "loss": 1.561,
+      "step": 1293
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019413909147454844,
+      "loss": 1.7394,
+      "step": 1294
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019412740982796613,
+      "loss": 1.2201,
+      "step": 1295
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019411571690351926,
+      "loss": 1.3653,
+      "step": 1296
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019410401270260878,
+      "loss": 1.7922,
+      "step": 1297
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019409229722663709,
+      "loss": 1.3396,
+      "step": 1298
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019408057047700785,
+      "loss": 1.1695,
+      "step": 1299
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019406883245512608,
+      "loss": 1.0802,
+      "step": 1300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001940570831623982,
+      "loss": 1.0679,
+      "step": 1301
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019404532260023193,
+      "loss": 1.6709,
+      "step": 1302
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019403355077003643,
+      "loss": 1.7189,
+      "step": 1303
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019402176767322203,
+      "loss": 1.1866,
+      "step": 1304
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019400997331120058,
+      "loss": 1.3205,
+      "step": 1305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001939981676853852,
+      "loss": 1.3797,
+      "step": 1306
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019398635079719042,
+      "loss": 1.4839,
+      "step": 1307
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019397452264803203,
+      "loss": 1.2429,
+      "step": 1308
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019396268323932725,
+      "loss": 1.2474,
+      "step": 1309
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001939508325724946,
+      "loss": 1.2851,
+      "step": 1310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019393897064895395,
+      "loss": 1.3249,
+      "step": 1311
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001939270974701266,
+      "loss": 1.2457,
+      "step": 1312
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001939152130374351,
+      "loss": 1.3405,
+      "step": 1313
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019390331735230333,
+      "loss": 1.4765,
+      "step": 1314
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019389141041615664,
+      "loss": 1.261,
+      "step": 1315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019387949223042164,
+      "loss": 1.5326,
+      "step": 1316
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019386756279652628,
+      "loss": 1.4923,
+      "step": 1317
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019385562211589989,
+      "loss": 1.4004,
+      "step": 1318
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019384367018997317,
+      "loss": 1.1965,
+      "step": 1319
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019383170702017813,
+      "loss": 1.193,
+      "step": 1320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001938197326079481,
+      "loss": 1.8224,
+      "step": 1321
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019380774695471783,
+      "loss": 1.3433,
+      "step": 1322
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001937957500619234,
+      "loss": 1.4716,
+      "step": 1323
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019378374193100215,
+      "loss": 1.428,
+      "step": 1324
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019377172256339288,
+      "loss": 1.0296,
+      "step": 1325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001937596919605357,
+      "loss": 1.2955,
+      "step": 1326
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000193747650123872,
+      "loss": 1.3569,
+      "step": 1327
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019373559705484464,
+      "loss": 1.238,
+      "step": 1328
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019372353275489768,
+      "loss": 1.3833,
+      "step": 1329
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019371145722547667,
+      "loss": 1.3226,
+      "step": 1330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019369937046802838,
+      "loss": 1.4167,
+      "step": 1331
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019368727248400105,
+      "loss": 1.1195,
+      "step": 1332
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019367516327484414,
+      "loss": 1.3814,
+      "step": 1333
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019366304284200857,
+      "loss": 1.6037,
+      "step": 1334
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001936509111869465,
+      "loss": 1.3146,
+      "step": 1335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019363876831111147,
+      "loss": 1.2328,
+      "step": 1336
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019362661421595844,
+      "loss": 1.2253,
+      "step": 1337
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019361444890294362,
+      "loss": 1.7427,
+      "step": 1338
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019360227237352458,
+      "loss": 1.3329,
+      "step": 1339
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019359008462916026,
+      "loss": 1.2413,
+      "step": 1340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019357788567131096,
+      "loss": 1.6039,
+      "step": 1341
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019356567550143824,
+      "loss": 1.3694,
+      "step": 1342
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001935534541210051,
+      "loss": 1.1974,
+      "step": 1343
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019354122153147584,
+      "loss": 1.5157,
+      "step": 1344
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001935289777343161,
+      "loss": 1.2594,
+      "step": 1345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019351672273099286,
+      "loss": 1.1844,
+      "step": 1346
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019350445652297448,
+      "loss": 1.0291,
+      "step": 1347
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019349217911173063,
+      "loss": 1.22,
+      "step": 1348
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001934798904987323,
+      "loss": 1.5519,
+      "step": 1349
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019346759068545187,
+      "loss": 1.1977,
+      "step": 1350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000193455279673363,
+      "loss": 1.2129,
+      "step": 1351
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019344295746394078,
+      "loss": 1.5817,
+      "step": 1352
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001934306240586616,
+      "loss": 1.2469,
+      "step": 1353
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019341827945900314,
+      "loss": 1.3315,
+      "step": 1354
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019340592366644453,
+      "loss": 1.3498,
+      "step": 1355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001933935566824661,
+      "loss": 1.5287,
+      "step": 1356
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019338117850854967,
+      "loss": 1.1264,
+      "step": 1357
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019336878914617828,
+      "loss": 1.9718,
+      "step": 1358
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019335638859683635,
+      "loss": 1.3139,
+      "step": 1359
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019334397686200972,
+      "loss": 1.1848,
+      "step": 1360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019333155394318544,
+      "loss": 1.1752,
+      "step": 1361
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019331911984185197,
+      "loss": 1.4146,
+      "step": 1362
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019330667455949916,
+      "loss": 1.0384,
+      "step": 1363
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019329421809761806,
+      "loss": 1.2188,
+      "step": 1364
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019328175045770114,
+      "loss": 1.3806,
+      "step": 1365
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019326927164124228,
+      "loss": 1.1605,
+      "step": 1366
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019325678164973655,
+      "loss": 1.3446,
+      "step": 1367
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001932442804846805,
+      "loss": 1.4541,
+      "step": 1368
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019323176814757192,
+      "loss": 1.51,
+      "step": 1369
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019321924463990998,
+      "loss": 1.4993,
+      "step": 1370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019320670996319521,
+      "loss": 1.4855,
+      "step": 1371
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019319416411892938,
+      "loss": 1.4849,
+      "step": 1372
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019318160710861575,
+      "loss": 1.4305,
+      "step": 1373
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019316903893375878,
+      "loss": 1.5434,
+      "step": 1374
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019315645959586437,
+      "loss": 1.6278,
+      "step": 1375
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019314386909643968,
+      "loss": 1.1396,
+      "step": 1376
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019313126743699324,
+      "loss": 1.6031,
+      "step": 1377
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019311865461903494,
+      "loss": 1.2212,
+      "step": 1378
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019310603064407594,
+      "loss": 1.3771,
+      "step": 1379
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019309339551362882,
+      "loss": 1.202,
+      "step": 1380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019308074922920744,
+      "loss": 1.3556,
+      "step": 1381
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000193068091792327,
+      "loss": 1.8367,
+      "step": 1382
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001930554232045041,
+      "loss": 1.389,
+      "step": 1383
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001930427434672566,
+      "loss": 1.222,
+      "step": 1384
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001930300525821037,
+      "loss": 1.096,
+      "step": 1385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019301735055056595,
+      "loss": 1.1955,
+      "step": 1386
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019300463737416528,
+      "loss": 1.4744,
+      "step": 1387
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019299191305442487,
+      "loss": 1.1438,
+      "step": 1388
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019297917759286934,
+      "loss": 1.4116,
+      "step": 1389
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019296643099102456,
+      "loss": 1.453,
+      "step": 1390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019295367325041776,
+      "loss": 1.3605,
+      "step": 1391
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001929409043725775,
+      "loss": 1.295,
+      "step": 1392
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001929281243590337,
+      "loss": 1.895,
+      "step": 1393
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001929153332113176,
+      "loss": 1.4736,
+      "step": 1394
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019290253093096174,
+      "loss": 1.16,
+      "step": 1395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019288971751950005,
+      "loss": 1.5453,
+      "step": 1396
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019287689297846776,
+      "loss": 1.4381,
+      "step": 1397
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019286405730940142,
+      "loss": 1.4102,
+      "step": 1398
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019285121051383896,
+      "loss": 1.7159,
+      "step": 1399
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019283835259331964,
+      "loss": 1.423,
+      "step": 1400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019282548354938398,
+      "loss": 1.0561,
+      "step": 1401
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001928126033835739,
+      "loss": 1.2634,
+      "step": 1402
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019279971209743265,
+      "loss": 1.6892,
+      "step": 1403
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019278680969250477,
+      "loss": 1.3952,
+      "step": 1404
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019277389617033621,
+      "loss": 1.5932,
+      "step": 1405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019276097153247414,
+      "loss": 1.412,
+      "step": 1406
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019274803578046718,
+      "loss": 1.2793,
+      "step": 1407
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019273508891586518,
+      "loss": 1.2789,
+      "step": 1408
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001927221309402194,
+      "loss": 1.2874,
+      "step": 1409
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019270916185508237,
+      "loss": 1.1927,
+      "step": 1410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000192696181662008,
+      "loss": 1.3111,
+      "step": 1411
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019268319036255149,
+      "loss": 1.1411,
+      "step": 1412
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019267018795826943,
+      "loss": 1.5968,
+      "step": 1413
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019265717445071968,
+      "loss": 1.0204,
+      "step": 1414
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019264414984146143,
+      "loss": 1.246,
+      "step": 1415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019263111413205523,
+      "loss": 1.0597,
+      "step": 1416
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000192618067324063,
+      "loss": 1.3742,
+      "step": 1417
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019260500941904786,
+      "loss": 1.8264,
+      "step": 1418
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019259194041857443,
+      "loss": 1.1021,
+      "step": 1419
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019257886032420852,
+      "loss": 1.306,
+      "step": 1420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001925657691375173,
+      "loss": 1.3877,
+      "step": 1421
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019255266686006933,
+      "loss": 1.2719,
+      "step": 1422
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019253955349343444,
+      "loss": 1.0655,
+      "step": 1423
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019252642903918384,
+      "loss": 1.5035,
+      "step": 1424
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019251329349889,
+      "loss": 1.7521,
+      "step": 1425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019250014687412673,
+      "loss": 1.3664,
+      "step": 1426
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001924869891664692,
+      "loss": 1.4185,
+      "step": 1427
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019247382037749396,
+      "loss": 1.2616,
+      "step": 1428
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019246064050877878,
+      "loss": 1.6248,
+      "step": 1429
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001924474495619028,
+      "loss": 1.5462,
+      "step": 1430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001924342475384465,
+      "loss": 1.3626,
+      "step": 1431
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001924210344399917,
+      "loss": 1.3884,
+      "step": 1432
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001924078102681215,
+      "loss": 1.2511,
+      "step": 1433
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019239457502442036,
+      "loss": 1.1466,
+      "step": 1434
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019238132871047406,
+      "loss": 1.4709,
+      "step": 1435
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001923680713278697,
+      "loss": 1.2987,
+      "step": 1436
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019235480287819573,
+      "loss": 1.3594,
+      "step": 1437
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019234152336304185,
+      "loss": 1.2699,
+      "step": 1438
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001923282327839992,
+      "loss": 1.3828,
+      "step": 1439
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001923149311426602,
+      "loss": 1.3154,
+      "step": 1440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019230161844061857,
+      "loss": 1.2571,
+      "step": 1441
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019228829467946934,
+      "loss": 1.2964,
+      "step": 1442
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001922749598608089,
+      "loss": 1.0458,
+      "step": 1443
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019226161398623498,
+      "loss": 1.2059,
+      "step": 1444
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019224825705734663,
+      "loss": 1.4604,
+      "step": 1445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019223488907574418,
+      "loss": 1.3344,
+      "step": 1446
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019222151004302934,
+      "loss": 1.3523,
+      "step": 1447
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019220811996080507,
+      "loss": 1.1737,
+      "step": 1448
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019219471883067575,
+      "loss": 1.3718,
+      "step": 1449
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000192181306654247,
+      "loss": 1.388,
+      "step": 1450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019216788343312584,
+      "loss": 1.3232,
+      "step": 1451
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019215444916892053,
+      "loss": 1.3504,
+      "step": 1452
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001921410038632407,
+      "loss": 1.5557,
+      "step": 1453
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001921275475176973,
+      "loss": 1.3791,
+      "step": 1454
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019211408013390262,
+      "loss": 1.392,
+      "step": 1455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019210060171347024,
+      "loss": 1.3638,
+      "step": 1456
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019208711225801508,
+      "loss": 1.4846,
+      "step": 1457
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019207361176915336,
+      "loss": 1.3148,
+      "step": 1458
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001920601002485027,
+      "loss": 1.6359,
+      "step": 1459
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019204657769768187,
+      "loss": 1.7119,
+      "step": 1460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019203304411831117,
+      "loss": 1.4778,
+      "step": 1461
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001920194995120121,
+      "loss": 1.6636,
+      "step": 1462
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019200594388040746,
+      "loss": 1.2199,
+      "step": 1463
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001919923772251215,
+      "loss": 1.3577,
+      "step": 1464
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019197879954777964,
+      "loss": 1.8794,
+      "step": 1465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001919652108500087,
+      "loss": 1.3818,
+      "step": 1466
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019195161113343682,
+      "loss": 1.4818,
+      "step": 1467
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019193800039969347,
+      "loss": 1.6232,
+      "step": 1468
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019192437865040937,
+      "loss": 1.0203,
+      "step": 1469
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019191074588721665,
+      "loss": 1.6252,
+      "step": 1470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019189710211174867,
+      "loss": 1.1652,
+      "step": 1471
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001918834473256402,
+      "loss": 1.2912,
+      "step": 1472
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019186978153052728,
+      "loss": 1.688,
+      "step": 1473
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019185610472804727,
+      "loss": 1.2616,
+      "step": 1474
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019184241691983883,
+      "loss": 1.3202,
+      "step": 1475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019182871810754202,
+      "loss": 1.0871,
+      "step": 1476
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019181500829279814,
+      "loss": 1.1754,
+      "step": 1477
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019180128747724976,
+      "loss": 1.1849,
+      "step": 1478
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019178755566254095,
+      "loss": 1.3092,
+      "step": 1479
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001917738128503169,
+      "loss": 1.1697,
+      "step": 1480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019176005904222426,
+      "loss": 1.2501,
+      "step": 1481
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001917462942399109,
+      "loss": 1.6872,
+      "step": 1482
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019173251844502607,
+      "loss": 1.4364,
+      "step": 1483
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001917187316592203,
+      "loss": 1.3165,
+      "step": 1484
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019170493388414548,
+      "loss": 1.1937,
+      "step": 1485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019169112512145477,
+      "loss": 1.4921,
+      "step": 1486
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019167730537280266,
+      "loss": 1.4858,
+      "step": 1487
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019166347463984496,
+      "loss": 1.2916,
+      "step": 1488
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001916496329242388,
+      "loss": 1.4759,
+      "step": 1489
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019163578022764264,
+      "loss": 1.3849,
+      "step": 1490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019162191655171625,
+      "loss": 0.956,
+      "step": 1491
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019160804189812065,
+      "loss": 1.3406,
+      "step": 1492
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019159415626851826,
+      "loss": 1.3235,
+      "step": 1493
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019158025966457284,
+      "loss": 1.5893,
+      "step": 1494
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001915663520879493,
+      "loss": 1.4845,
+      "step": 1495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019155243354031405,
+      "loss": 1.2434,
+      "step": 1496
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019153850402333473,
+      "loss": 1.3286,
+      "step": 1497
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019152456353868028,
+      "loss": 1.3453,
+      "step": 1498
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019151061208802102,
+      "loss": 1.2709,
+      "step": 1499
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019149664967302847,
+      "loss": 1.6029,
+      "step": 1500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001914826762953756,
+      "loss": 1.1954,
+      "step": 1501
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019146869195673662,
+      "loss": 1.537,
+      "step": 1502
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019145469665878703,
+      "loss": 1.1591,
+      "step": 1503
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019144069040320373,
+      "loss": 1.104,
+      "step": 1504
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019142667319166484,
+      "loss": 1.3171,
+      "step": 1505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019141264502584978,
+      "loss": 1.1631,
+      "step": 1506
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019139860590743944,
+      "loss": 1.3295,
+      "step": 1507
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019138455583811583,
+      "loss": 1.2227,
+      "step": 1508
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001913704948195624,
+      "loss": 1.2015,
+      "step": 1509
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019135642285346383,
+      "loss": 1.1688,
+      "step": 1510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019134233994150623,
+      "loss": 1.3056,
+      "step": 1511
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019132824608537687,
+      "loss": 1.3668,
+      "step": 1512
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019131414128676442,
+      "loss": 1.173,
+      "step": 1513
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019130002554735884,
+      "loss": 1.3654,
+      "step": 1514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019128589886885144,
+      "loss": 1.1313,
+      "step": 1515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019127176125293476,
+      "loss": 1.3004,
+      "step": 1516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019125761270130274,
+      "loss": 1.1715,
+      "step": 1517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019124345321565054,
+      "loss": 1.4637,
+      "step": 1518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019122928279767474,
+      "loss": 1.3287,
+      "step": 1519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001912151014490731,
+      "loss": 1.3995,
+      "step": 1520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001912009091715448,
+      "loss": 1.3444,
+      "step": 1521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019118670596679027,
+      "loss": 1.2986,
+      "step": 1522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019117249183651125,
+      "loss": 1.4719,
+      "step": 1523
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019115826678241083,
+      "loss": 1.5916,
+      "step": 1524
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001911440308061934,
+      "loss": 1.3927,
+      "step": 1525
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001911297839095646,
+      "loss": 1.3307,
+      "step": 1526
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019111552609423144,
+      "loss": 1.3228,
+      "step": 1527
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019110125736190221,
+      "loss": 1.4524,
+      "step": 1528
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019108697771428652,
+      "loss": 1.1754,
+      "step": 1529
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001910726871530953,
+      "loss": 1.3398,
+      "step": 1530
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019105838568004075,
+      "loss": 1.5885,
+      "step": 1531
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019104407329683642,
+      "loss": 1.1587,
+      "step": 1532
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001910297500051971,
+      "loss": 1.4734,
+      "step": 1533
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000191015415806839,
+      "loss": 1.4515,
+      "step": 1534
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019100107070347955,
+      "loss": 1.1829,
+      "step": 1535
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019098671469683747,
+      "loss": 1.3269,
+      "step": 1536
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019097234778863285,
+      "loss": 1.4442,
+      "step": 1537
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019095796998058707,
+      "loss": 1.2487,
+      "step": 1538
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019094358127442278,
+      "loss": 1.0703,
+      "step": 1539
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019092918167186397,
+      "loss": 1.1228,
+      "step": 1540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019091477117463593,
+      "loss": 1.2275,
+      "step": 1541
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019090034978446529,
+      "loss": 1.2813,
+      "step": 1542
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019088591750307987,
+      "loss": 1.3075,
+      "step": 1543
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019087147433220894,
+      "loss": 1.4149,
+      "step": 1544
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019085702027358298,
+      "loss": 1.1174,
+      "step": 1545
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019084255532893376,
+      "loss": 1.1389,
+      "step": 1546
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019082807949999446,
+      "loss": 1.4284,
+      "step": 1547
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019081359278849948,
+      "loss": 1.3021,
+      "step": 1548
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019079909519618453,
+      "loss": 1.3501,
+      "step": 1549
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019078458672478667,
+      "loss": 1.0589,
+      "step": 1550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001907700673760442,
+      "loss": 1.322,
+      "step": 1551
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019075553715169674,
+      "loss": 1.8916,
+      "step": 1552
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019074099605348525,
+      "loss": 1.2412,
+      "step": 1553
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019072644408315198,
+      "loss": 1.4838,
+      "step": 1554
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001907118812424405,
+      "loss": 1.43,
+      "step": 1555
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019069730753309557,
+      "loss": 1.6426,
+      "step": 1556
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001906827229568634,
+      "loss": 1.3219,
+      "step": 1557
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001906681275154914,
+      "loss": 1.6204,
+      "step": 1558
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019065352121072838,
+      "loss": 1.144,
+      "step": 1559
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019063890404432435,
+      "loss": 1.6413,
+      "step": 1560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019062427601803064,
+      "loss": 1.3191,
+      "step": 1561
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019060963713359995,
+      "loss": 1.5307,
+      "step": 1562
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019059498739278622,
+      "loss": 1.5617,
+      "step": 1563
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019058032679734473,
+      "loss": 1.4036,
+      "step": 1564
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000190565655349032,
+      "loss": 1.0932,
+      "step": 1565
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019055097304960592,
+      "loss": 1.1525,
+      "step": 1566
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001905362799008256,
+      "loss": 0.9018,
+      "step": 1567
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019052157590445157,
+      "loss": 1.1424,
+      "step": 1568
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019050686106224552,
+      "loss": 1.1553,
+      "step": 1569
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019049213537597056,
+      "loss": 1.4044,
+      "step": 1570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000190477398847391,
+      "loss": 1.4477,
+      "step": 1571
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019046265147827253,
+      "loss": 0.8492,
+      "step": 1572
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001904478932703821,
+      "loss": 1.5042,
+      "step": 1573
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019043312422548794,
+      "loss": 1.7656,
+      "step": 1574
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019041834434535958,
+      "loss": 1.0601,
+      "step": 1575
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019040355363176797,
+      "loss": 1.2271,
+      "step": 1576
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019038875208648515,
+      "loss": 1.1974,
+      "step": 1577
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019037393971128465,
+      "loss": 1.3383,
+      "step": 1578
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019035911650794112,
+      "loss": 1.1633,
+      "step": 1579
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019034428247823067,
+      "loss": 1.5612,
+      "step": 1580
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019032943762393062,
+      "loss": 1.1789,
+      "step": 1581
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019031458194681964,
+      "loss": 1.1915,
+      "step": 1582
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001902997154486776,
+      "loss": 1.7192,
+      "step": 1583
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019028483813128574,
+      "loss": 1.1619,
+      "step": 1584
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019026994999642662,
+      "loss": 1.1505,
+      "step": 1585
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000190255051045884,
+      "loss": 1.3891,
+      "step": 1586
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019024014128144306,
+      "loss": 1.0376,
+      "step": 1587
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019022522070489018,
+      "loss": 1.1786,
+      "step": 1588
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019021028931801305,
+      "loss": 1.1301,
+      "step": 1589
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019019534712260076,
+      "loss": 1.7146,
+      "step": 1590
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019018039412044347,
+      "loss": 1.3696,
+      "step": 1591
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019016543031333288,
+      "loss": 1.4517,
+      "step": 1592
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019015045570306183,
+      "loss": 1.3754,
+      "step": 1593
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001901354702914245,
+      "loss": 1.6473,
+      "step": 1594
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001901204740802164,
+      "loss": 1.5013,
+      "step": 1595
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019010546707123424,
+      "loss": 1.3609,
+      "step": 1596
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019009044926627614,
+      "loss": 1.4233,
+      "step": 1597
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019007542066714147,
+      "loss": 1.5343,
+      "step": 1598
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001900603812756308,
+      "loss": 1.5389,
+      "step": 1599
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019004533109354618,
+      "loss": 1.5431,
+      "step": 1600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019003027012269073,
+      "loss": 1.7175,
+      "step": 1601
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019001519836486902,
+      "loss": 1.1311,
+      "step": 1602
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001900001158218869,
+      "loss": 1.3398,
+      "step": 1603
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018998502249555147,
+      "loss": 1.2439,
+      "step": 1604
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018996991838767112,
+      "loss": 1.5745,
+      "step": 1605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001899548035000556,
+      "loss": 1.4831,
+      "step": 1606
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018993967783451584,
+      "loss": 1.0406,
+      "step": 1607
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018992454139286414,
+      "loss": 1.2405,
+      "step": 1608
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018990939417691402,
+      "loss": 1.3574,
+      "step": 1609
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018989423618848044,
+      "loss": 1.1074,
+      "step": 1610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001898790674293795,
+      "loss": 1.3041,
+      "step": 1611
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018986388790142865,
+      "loss": 1.2074,
+      "step": 1612
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001898486976064466,
+      "loss": 1.6856,
+      "step": 1613
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018983349654625337,
+      "loss": 1.2862,
+      "step": 1614
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018981828472267034,
+      "loss": 1.4651,
+      "step": 1615
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018980306213752004,
+      "loss": 1.5008,
+      "step": 1616
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001897878287926264,
+      "loss": 1.3614,
+      "step": 1617
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018977258468981458,
+      "loss": 1.3431,
+      "step": 1618
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018975732983091107,
+      "loss": 1.6818,
+      "step": 1619
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018974206421774366,
+      "loss": 1.2828,
+      "step": 1620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018972678785214133,
+      "loss": 0.993,
+      "step": 1621
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018971150073593445,
+      "loss": 1.5419,
+      "step": 1622
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018969620287095462,
+      "loss": 1.4938,
+      "step": 1623
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018968089425903476,
+      "loss": 1.3535,
+      "step": 1624
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018966557490200913,
+      "loss": 1.204,
+      "step": 1625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018965024480171314,
+      "loss": 1.1914,
+      "step": 1626
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001896349039599836,
+      "loss": 1.2904,
+      "step": 1627
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018961955237865855,
+      "loss": 1.2884,
+      "step": 1628
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018960419005957738,
+      "loss": 1.2582,
+      "step": 1629
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018958881700458066,
+      "loss": 1.432,
+      "step": 1630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018957343321551038,
+      "loss": 1.1958,
+      "step": 1631
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001895580386942097,
+      "loss": 1.5635,
+      "step": 1632
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001895426334425231,
+      "loss": 1.5865,
+      "step": 1633
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001895272174622964,
+      "loss": 1.8818,
+      "step": 1634
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001895117907553767,
+      "loss": 1.0378,
+      "step": 1635
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018949635332361222,
+      "loss": 1.5573,
+      "step": 1636
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001894809051688527,
+      "loss": 1.4176,
+      "step": 1637
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018946544629294904,
+      "loss": 1.1002,
+      "step": 1638
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018944997669775342,
+      "loss": 1.3483,
+      "step": 1639
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018943449638511933,
+      "loss": 1.235,
+      "step": 1640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001894190053569016,
+      "loss": 1.8071,
+      "step": 1641
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001894035036149562,
+      "loss": 1.5054,
+      "step": 1642
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018938799116114053,
+      "loss": 1.3114,
+      "step": 1643
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018937246799731317,
+      "loss": 1.1753,
+      "step": 1644
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018935693412533406,
+      "loss": 1.0857,
+      "step": 1645
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018934138954706438,
+      "loss": 1.2798,
+      "step": 1646
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018932583426436658,
+      "loss": 0.9648,
+      "step": 1647
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018931026827910446,
+      "loss": 1.3813,
+      "step": 1648
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000189294691593143,
+      "loss": 1.4989,
+      "step": 1649
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001892791042083486,
+      "loss": 1.7211,
+      "step": 1650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018926350612658875,
+      "loss": 1.3627,
+      "step": 1651
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018924789734973242,
+      "loss": 1.4924,
+      "step": 1652
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018923227787964972,
+      "loss": 1.1902,
+      "step": 1653
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018921664771821215,
+      "loss": 1.3727,
+      "step": 1654
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018920100686729239,
+      "loss": 1.2201,
+      "step": 1655
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018918535532876446,
+      "loss": 1.5097,
+      "step": 1656
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018916969310450366,
+      "loss": 1.1721,
+      "step": 1657
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018915402019638653,
+      "loss": 1.2635,
+      "step": 1658
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018913833660629095,
+      "loss": 1.2562,
+      "step": 1659
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018912264233609603,
+      "loss": 1.1245,
+      "step": 1660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018910693738768218,
+      "loss": 1.3359,
+      "step": 1661
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018909122176293108,
+      "loss": 1.2296,
+      "step": 1662
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001890754954637257,
+      "loss": 0.9474,
+      "step": 1663
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001890597584919503,
+      "loss": 1.3539,
+      "step": 1664
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001890440108494904,
+      "loss": 1.2429,
+      "step": 1665
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018902825253823275,
+      "loss": 1.1051,
+      "step": 1666
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001890124835600655,
+      "loss": 1.4207,
+      "step": 1667
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018899670391687797,
+      "loss": 1.7097,
+      "step": 1668
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018898091361056082,
+      "loss": 1.398,
+      "step": 1669
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018896511264300596,
+      "loss": 1.3824,
+      "step": 1670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018894930101610657,
+      "loss": 1.5604,
+      "step": 1671
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001889334787317571,
+      "loss": 1.5165,
+      "step": 1672
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018891764579185332,
+      "loss": 1.335,
+      "step": 1673
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018890180219829222,
+      "loss": 1.1637,
+      "step": 1674
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018888594795297216,
+      "loss": 1.4765,
+      "step": 1675
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018887008305779268,
+      "loss": 1.3151,
+      "step": 1676
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018885420751465462,
+      "loss": 1.4481,
+      "step": 1677
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018883832132546012,
+      "loss": 1.272,
+      "step": 1678
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018882242449211258,
+      "loss": 1.438,
+      "step": 1679
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018880651701651667,
+      "loss": 1.1824,
+      "step": 1680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018879059890057834,
+      "loss": 1.5795,
+      "step": 1681
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018877467014620484,
+      "loss": 1.3468,
+      "step": 1682
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018875873075530463,
+      "loss": 1.4286,
+      "step": 1683
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018874278072978756,
+      "loss": 1.0602,
+      "step": 1684
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018872682007156457,
+      "loss": 1.2967,
+      "step": 1685
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001887108487825481,
+      "loss": 1.4077,
+      "step": 1686
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018869486686465167,
+      "loss": 1.5897,
+      "step": 1687
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001886788743197902,
+      "loss": 1.3123,
+      "step": 1688
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001886628711498798,
+      "loss": 1.0687,
+      "step": 1689
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001886468573568379,
+      "loss": 1.8219,
+      "step": 1690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018863083294258317,
+      "loss": 1.3729,
+      "step": 1691
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018861479790903562,
+      "loss": 1.3673,
+      "step": 1692
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018859875225811645,
+      "loss": 1.5276,
+      "step": 1693
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001885826959917482,
+      "loss": 1.3901,
+      "step": 1694
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018856662911185458,
+      "loss": 1.321,
+      "step": 1695
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018855055162036072,
+      "loss": 1.1994,
+      "step": 1696
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018853446351919294,
+      "loss": 1.3269,
+      "step": 1697
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018851836481027876,
+      "loss": 1.2834,
+      "step": 1698
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001885022554955471,
+      "loss": 1.1858,
+      "step": 1699
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001884861355769281,
+      "loss": 1.5928,
+      "step": 1700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018847000505635314,
+      "loss": 1.2035,
+      "step": 1701
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018845386393575493,
+      "loss": 1.4789,
+      "step": 1702
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018843771221706734,
+      "loss": 1.345,
+      "step": 1703
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001884215499022257,
+      "loss": 1.3503,
+      "step": 1704
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018840537699316642,
+      "loss": 1.1951,
+      "step": 1705
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018838919349182726,
+      "loss": 1.2089,
+      "step": 1706
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018837299940014727,
+      "loss": 0.9982,
+      "step": 1707
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001883567947200667,
+      "loss": 1.6106,
+      "step": 1708
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018834057945352718,
+      "loss": 1.4429,
+      "step": 1709
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001883243536024715,
+      "loss": 1.2595,
+      "step": 1710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018830811716884376,
+      "loss": 1.2666,
+      "step": 1711
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018829187015458932,
+      "loss": 1.3724,
+      "step": 1712
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001882756125616548,
+      "loss": 1.0892,
+      "step": 1713
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018825934439198813,
+      "loss": 1.2385,
+      "step": 1714
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018824306564753848,
+      "loss": 1.2606,
+      "step": 1715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001882267763302563,
+      "loss": 1.083,
+      "step": 1716
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018821047644209323,
+      "loss": 1.3205,
+      "step": 1717
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001881941659850023,
+      "loss": 1.5178,
+      "step": 1718
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001881778449609377,
+      "loss": 1.45,
+      "step": 1719
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000188161513371855,
+      "loss": 1.3624,
+      "step": 1720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001881451712197109,
+      "loss": 1.323,
+      "step": 1721
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018812881850646348,
+      "loss": 1.1312,
+      "step": 1722
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018811245523407197,
+      "loss": 0.9858,
+      "step": 1723
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000188096081404497,
+      "loss": 1.6752,
+      "step": 1724
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018807969701970038,
+      "loss": 1.3198,
+      "step": 1725
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001880633020816452,
+      "loss": 1.3633,
+      "step": 1726
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018804689659229583,
+      "loss": 1.5468,
+      "step": 1727
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018803048055361784,
+      "loss": 1.3517,
+      "step": 1728
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001880140539675782,
+      "loss": 1.453,
+      "step": 1729
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000187997616836145,
+      "loss": 1.1973,
+      "step": 1730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018798116916128767,
+      "loss": 1.5012,
+      "step": 1731
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018796471094497687,
+      "loss": 1.2722,
+      "step": 1732
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018794824218918457,
+      "loss": 1.4809,
+      "step": 1733
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018793176289588396,
+      "loss": 1.6252,
+      "step": 1734
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018791527306704947,
+      "loss": 1.5725,
+      "step": 1735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001878987727046569,
+      "loss": 1.2637,
+      "step": 1736
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018788226181068316,
+      "loss": 1.656,
+      "step": 1737
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018786574038710653,
+      "loss": 1.5001,
+      "step": 1738
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018784920843590655,
+      "loss": 1.5219,
+      "step": 1739
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018783266595906396,
+      "loss": 1.4864,
+      "step": 1740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018781611295856082,
+      "loss": 1.1619,
+      "step": 1741
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001877995494363804,
+      "loss": 1.6648,
+      "step": 1742
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001877829753945073,
+      "loss": 1.5011,
+      "step": 1743
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001877663908349273,
+      "loss": 1.7013,
+      "step": 1744
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018774979575962746,
+      "loss": 1.4738,
+      "step": 1745
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018773319017059613,
+      "loss": 1.1948,
+      "step": 1746
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018771657406982297,
+      "loss": 1.342,
+      "step": 1747
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018769994745929873,
+      "loss": 1.3173,
+      "step": 1748
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001876833103410156,
+      "loss": 1.4662,
+      "step": 1749
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018766666271696693,
+      "loss": 1.3626,
+      "step": 1750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018765000458914737,
+      "loss": 1.4025,
+      "step": 1751
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001876333359595528,
+      "loss": 1.4444,
+      "step": 1752
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018761665683018035,
+      "loss": 1.4343,
+      "step": 1753
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018759996720302846,
+      "loss": 1.4009,
+      "step": 1754
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018758326708009676,
+      "loss": 1.2191,
+      "step": 1755
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018756655646338623,
+      "loss": 1.1557,
+      "step": 1756
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000187549835354899,
+      "loss": 1.4759,
+      "step": 1757
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018753310375663852,
+      "loss": 1.2074,
+      "step": 1758
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001875163616706095,
+      "loss": 1.319,
+      "step": 1759
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018749960909881787,
+      "loss": 1.3001,
+      "step": 1760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018748284604327083,
+      "loss": 1.4386,
+      "step": 1761
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018746607250597687,
+      "loss": 1.6137,
+      "step": 1762
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018744928848894572,
+      "loss": 1.4022,
+      "step": 1763
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018743249399418832,
+      "loss": 1.2797,
+      "step": 1764
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001874156890237169,
+      "loss": 1.4986,
+      "step": 1765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018739887357954496,
+      "loss": 1.2129,
+      "step": 1766
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018738204766368728,
+      "loss": 1.3173,
+      "step": 1767
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018736521127815978,
+      "loss": 1.2949,
+      "step": 1768
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018734836442497977,
+      "loss": 1.2832,
+      "step": 1769
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001873315071061657,
+      "loss": 1.2518,
+      "step": 1770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018731463932373737,
+      "loss": 1.1302,
+      "step": 1771
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001872977610797158,
+      "loss": 1.7315,
+      "step": 1772
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001872808723761232,
+      "loss": 1.3878,
+      "step": 1773
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018726397321498314,
+      "loss": 1.3233,
+      "step": 1774
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018724706359832038,
+      "loss": 0.9443,
+      "step": 1775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001872301435281609,
+      "loss": 1.6932,
+      "step": 1776
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018721321300653205,
+      "loss": 1.173,
+      "step": 1777
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018719627203546231,
+      "loss": 1.1697,
+      "step": 1778
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018717932061698144,
+      "loss": 1.7326,
+      "step": 1779
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018716235875312055,
+      "loss": 1.2883,
+      "step": 1780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018714538644591184,
+      "loss": 1.2975,
+      "step": 1781
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018712840369738891,
+      "loss": 1.465,
+      "step": 1782
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001871114105095865,
+      "loss": 1.5262,
+      "step": 1783
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001870944068845407,
+      "loss": 0.9107,
+      "step": 1784
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001870773928242887,
+      "loss": 1.2518,
+      "step": 1785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018706036833086916,
+      "loss": 1.3571,
+      "step": 1786
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018704333340632177,
+      "loss": 1.0623,
+      "step": 1787
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001870262880526876,
+      "loss": 0.998,
+      "step": 1788
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018700923227200897,
+      "loss": 1.0918,
+      "step": 1789
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018699216606632936,
+      "loss": 1.5065,
+      "step": 1790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018697508943769358,
+      "loss": 1.6184,
+      "step": 1791
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018695800238814774,
+      "loss": 1.4377,
+      "step": 1792
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000186940904919739,
+      "loss": 1.3794,
+      "step": 1793
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018692379703451591,
+      "loss": 1.4961,
+      "step": 1794
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018690667873452833,
+      "loss": 1.4062,
+      "step": 1795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001868895500218272,
+      "loss": 1.541,
+      "step": 1796
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018687241089846487,
+      "loss": 1.4441,
+      "step": 1797
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001868552613664948,
+      "loss": 1.0733,
+      "step": 1798
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001868381014279718,
+      "loss": 1.1526,
+      "step": 1799
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018682093108495185,
+      "loss": 1.2093,
+      "step": 1800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018680375033949223,
+      "loss": 1.6246,
+      "step": 1801
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018678655919365152,
+      "loss": 1.0657,
+      "step": 1802
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018676935764948933,
+      "loss": 1.7909,
+      "step": 1803
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018675214570906676,
+      "loss": 1.5284,
+      "step": 1804
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018673492337444603,
+      "loss": 1.5257,
+      "step": 1805
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018671769064769066,
+      "loss": 1.3888,
+      "step": 1806
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018670044753086535,
+      "loss": 1.3809,
+      "step": 1807
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018668319402603606,
+      "loss": 1.6958,
+      "step": 1808
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001866659301352701,
+      "loss": 1.2807,
+      "step": 1809
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018664865586063588,
+      "loss": 1.2422,
+      "step": 1810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018663137120420315,
+      "loss": 1.1361,
+      "step": 1811
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018661407616804283,
+      "loss": 1.2783,
+      "step": 1812
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018659677075422713,
+      "loss": 1.3,
+      "step": 1813
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001865794549648295,
+      "loss": 1.2258,
+      "step": 1814
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018656212880192465,
+      "loss": 1.5358,
+      "step": 1815
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018654479226758852,
+      "loss": 1.3473,
+      "step": 1816
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018652744536389824,
+      "loss": 1.3853,
+      "step": 1817
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018651008809293226,
+      "loss": 1.0484,
+      "step": 1818
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018649272045677019,
+      "loss": 1.4547,
+      "step": 1819
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018647534245749302,
+      "loss": 1.0001,
+      "step": 1820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018645795409718283,
+      "loss": 1.4593,
+      "step": 1821
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000186440555377923,
+      "loss": 1.4954,
+      "step": 1822
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001864231463017982,
+      "loss": 1.3956,
+      "step": 1823
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018640572687089423,
+      "loss": 1.5717,
+      "step": 1824
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001863882970872983,
+      "loss": 1.1535,
+      "step": 1825
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018637085695309864,
+      "loss": 1.2585,
+      "step": 1826
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001863534064703849,
+      "loss": 1.4147,
+      "step": 1827
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001863359456412479,
+      "loss": 1.4068,
+      "step": 1828
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018631847446777972,
+      "loss": 1.4565,
+      "step": 1829
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018630099295207365,
+      "loss": 1.2429,
+      "step": 1830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018628350109622425,
+      "loss": 1.3753,
+      "step": 1831
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001862659989023273,
+      "loss": 1.3989,
+      "step": 1832
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018624848637247978,
+      "loss": 1.2784,
+      "step": 1833
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018623096350878001,
+      "loss": 1.2057,
+      "step": 1834
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018621343031332747,
+      "loss": 1.0801,
+      "step": 1835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001861958867882229,
+      "loss": 1.5208,
+      "step": 1836
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018617833293556828,
+      "loss": 1.4161,
+      "step": 1837
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001861607687574668,
+      "loss": 1.1151,
+      "step": 1838
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018614319425602294,
+      "loss": 1.529,
+      "step": 1839
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018612560943334234,
+      "loss": 1.6152,
+      "step": 1840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018610801429153198,
+      "loss": 1.3936,
+      "step": 1841
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018609040883269998,
+      "loss": 1.5486,
+      "step": 1842
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018607279305895576,
+      "loss": 1.4005,
+      "step": 1843
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018605516697240993,
+      "loss": 1.4823,
+      "step": 1844
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018603753057517438,
+      "loss": 1.2437,
+      "step": 1845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018601988386936222,
+      "loss": 1.3214,
+      "step": 1846
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018600222685708777,
+      "loss": 1.3636,
+      "step": 1847
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018598455954046657,
+      "loss": 1.0057,
+      "step": 1848
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018596688192161552,
+      "loss": 1.687,
+      "step": 1849
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018594919400265254,
+      "loss": 1.2581,
+      "step": 1850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000185931495785697,
+      "loss": 1.3781,
+      "step": 1851
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018591378727286935,
+      "loss": 1.5226,
+      "step": 1852
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018589606846629143,
+      "loss": 1.5988,
+      "step": 1853
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018587833936808607,
+      "loss": 1.3049,
+      "step": 1854
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018586059998037762,
+      "loss": 1.538,
+      "step": 1855
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018584285030529143,
+      "loss": 1.3227,
+      "step": 1856
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018582509034495418,
+      "loss": 1.3956,
+      "step": 1857
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018580732010149386,
+      "loss": 0.9954,
+      "step": 1858
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018578953957703952,
+      "loss": 1.6344,
+      "step": 1859
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001857717487737216,
+      "loss": 1.2592,
+      "step": 1860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018575394769367165,
+      "loss": 1.0644,
+      "step": 1861
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018573613633902252,
+      "loss": 1.5535,
+      "step": 1862
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018571831471190823,
+      "loss": 1.4142,
+      "step": 1863
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001857004828144642,
+      "loss": 1.583,
+      "step": 1864
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018568264064882682,
+      "loss": 1.1671,
+      "step": 1865
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018566478821713397,
+      "loss": 1.1154,
+      "step": 1866
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018564692552152452,
+      "loss": 1.3509,
+      "step": 1867
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018562905256413879,
+      "loss": 1.3986,
+      "step": 1868
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018561116934711814,
+      "loss": 1.3986,
+      "step": 1869
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001855932758726053,
+      "loss": 1.1968,
+      "step": 1870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018557537214274414,
+      "loss": 1.1478,
+      "step": 1871
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018555745815967983,
+      "loss": 1.2809,
+      "step": 1872
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018553953392555871,
+      "loss": 1.3544,
+      "step": 1873
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018552159944252837,
+      "loss": 1.4221,
+      "step": 1874
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018550365471273764,
+      "loss": 1.2625,
+      "step": 1875
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018548569973833657,
+      "loss": 1.2278,
+      "step": 1876
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001854677345214764,
+      "loss": 1.5513,
+      "step": 1877
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018544975906430964,
+      "loss": 1.219,
+      "step": 1878
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018543177336899006,
+      "loss": 1.4857,
+      "step": 1879
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018541377743767253,
+      "loss": 1.1705,
+      "step": 1880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001853957712725133,
+      "loss": 1.1506,
+      "step": 1881
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018537775487566978,
+      "loss": 1.4361,
+      "step": 1882
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018535972824930056,
+      "loss": 0.8798,
+      "step": 1883
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018534169139556556,
+      "loss": 1.2024,
+      "step": 1884
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018532364431662578,
+      "loss": 1.1799,
+      "step": 1885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018530558701464357,
+      "loss": 1.1562,
+      "step": 1886
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001852875194917825,
+      "loss": 1.2881,
+      "step": 1887
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018526944175020727,
+      "loss": 1.2148,
+      "step": 1888
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001852513537920839,
+      "loss": 1.4923,
+      "step": 1889
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018523325561957955,
+      "loss": 1.1669,
+      "step": 1890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001852151472348627,
+      "loss": 1.722,
+      "step": 1891
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018519702864010302,
+      "loss": 1.2167,
+      "step": 1892
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001851788998374713,
+      "loss": 1.1778,
+      "step": 1893
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018516076082913973,
+      "loss": 1.6037,
+      "step": 1894
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001851426116172816,
+      "loss": 1.4098,
+      "step": 1895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018512445220407141,
+      "loss": 0.964,
+      "step": 1896
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018510628259168498,
+      "loss": 1.2741,
+      "step": 1897
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018508810278229933,
+      "loss": 1.3175,
+      "step": 1898
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001850699127780926,
+      "loss": 1.3691,
+      "step": 1899
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018505171258124427,
+      "loss": 1.4769,
+      "step": 1900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018503350219393498,
+      "loss": 1.2787,
+      "step": 1901
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018501528161834663,
+      "loss": 1.1084,
+      "step": 1902
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018499705085666226,
+      "loss": 1.5538,
+      "step": 1903
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018497880991106622,
+      "loss": 1.0653,
+      "step": 1904
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018496055878374407,
+      "loss": 0.9551,
+      "step": 1905
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001849422974768825,
+      "loss": 1.3891,
+      "step": 1906
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018492402599266954,
+      "loss": 0.9634,
+      "step": 1907
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001849057443332944,
+      "loss": 1.2725,
+      "step": 1908
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018488745250094745,
+      "loss": 1.3596,
+      "step": 1909
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018486915049782034,
+      "loss": 1.1677,
+      "step": 1910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018485083832610593,
+      "loss": 1.2475,
+      "step": 1911
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018483251598799827,
+      "loss": 1.245,
+      "step": 1912
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018481418348569267,
+      "loss": 1.2143,
+      "step": 1913
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018479584082138568,
+      "loss": 1.4377,
+      "step": 1914
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018477748799727492,
+      "loss": 1.2236,
+      "step": 1915
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001847591250155594,
+      "loss": 1.4512,
+      "step": 1916
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018474075187843928,
+      "loss": 1.2264,
+      "step": 1917
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001847223685881159,
+      "loss": 1.0908,
+      "step": 1918
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018470397514679188,
+      "loss": 1.1815,
+      "step": 1919
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018468557155667102,
+      "loss": 1.2502,
+      "step": 1920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018466715781995837,
+      "loss": 1.5344,
+      "step": 1921
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018464873393886008,
+      "loss": 1.1527,
+      "step": 1922
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018463029991558371,
+      "loss": 1.2852,
+      "step": 1923
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018461185575233789,
+      "loss": 1.3477,
+      "step": 1924
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001845934014513325,
+      "loss": 1.759,
+      "step": 1925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018457493701477866,
+      "loss": 1.3576,
+      "step": 1926
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018455646244488864,
+      "loss": 1.6536,
+      "step": 1927
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018453797774387601,
+      "loss": 1.2714,
+      "step": 1928
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001845194829139555,
+      "loss": 1.4809,
+      "step": 1929
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001845009779573431,
+      "loss": 1.3983,
+      "step": 1930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001844824628762559,
+      "loss": 1.1186,
+      "step": 1931
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018446393767291232,
+      "loss": 1.0436,
+      "step": 1932
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000184445402349532,
+      "loss": 1.4895,
+      "step": 1933
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018442685690833567,
+      "loss": 1.6343,
+      "step": 1934
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018440830135154542,
+      "loss": 1.2807,
+      "step": 1935
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018438973568138444,
+      "loss": 1.2334,
+      "step": 1936
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001843711599000772,
+      "loss": 1.6232,
+      "step": 1937
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018435257400984932,
+      "loss": 1.1922,
+      "step": 1938
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018433397801292765,
+      "loss": 1.6855,
+      "step": 1939
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018431537191154038,
+      "loss": 1.5923,
+      "step": 1940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018429675570791668,
+      "loss": 1.526,
+      "step": 1941
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001842781294042871,
+      "loss": 1.3139,
+      "step": 1942
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018425949300288332,
+      "loss": 1.2829,
+      "step": 1943
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001842408465059383,
+      "loss": 1.429,
+      "step": 1944
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018422218991568615,
+      "loss": 1.6064,
+      "step": 1945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001842035232343622,
+      "loss": 1.2507,
+      "step": 1946
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018418484646420295,
+      "loss": 1.1442,
+      "step": 1947
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018416615960744622,
+      "loss": 1.4357,
+      "step": 1948
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000184147462666331,
+      "loss": 1.5037,
+      "step": 1949
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018412875564309737,
+      "loss": 1.3275,
+      "step": 1950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001841100385399868,
+      "loss": 1.4072,
+      "step": 1951
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018409131135924177,
+      "loss": 1.1895,
+      "step": 1952
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001840725741031062,
+      "loss": 1.3334,
+      "step": 1953
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018405382677382502,
+      "loss": 1.6656,
+      "step": 1954
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018403506937364442,
+      "loss": 1.2414,
+      "step": 1955
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001840163019048119,
+      "loss": 0.939,
+      "step": 1956
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000183997524369576,
+      "loss": 1.196,
+      "step": 1957
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001839787367701866,
+      "loss": 1.3683,
+      "step": 1958
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018395993910889463,
+      "loss": 1.4213,
+      "step": 1959
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018394113138795249,
+      "loss": 1.0536,
+      "step": 1960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018392231360961353,
+      "loss": 1.6928,
+      "step": 1961
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018390348577613242,
+      "loss": 1.4496,
+      "step": 1962
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000183884647889765,
+      "loss": 1.1847,
+      "step": 1963
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018386579995276835,
+      "loss": 1.7288,
+      "step": 1964
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018384694196740072,
+      "loss": 1.2824,
+      "step": 1965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018382807393592154,
+      "loss": 1.2603,
+      "step": 1966
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018380919586059155,
+      "loss": 1.1234,
+      "step": 1967
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018379030774367261,
+      "loss": 1.2971,
+      "step": 1968
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018377140958742775,
+      "loss": 1.3276,
+      "step": 1969
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018375250139412132,
+      "loss": 1.5043,
+      "step": 1970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018373358316601875,
+      "loss": 1.4074,
+      "step": 1971
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018371465490538675,
+      "loss": 1.4166,
+      "step": 1972
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001836957166144932,
+      "loss": 1.1352,
+      "step": 1973
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018367676829560717,
+      "loss": 1.4157,
+      "step": 1974
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018365780995099898,
+      "loss": 1.2539,
+      "step": 1975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001836388415829401,
+      "loss": 1.4337,
+      "step": 1976
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018361986319370325,
+      "loss": 1.4787,
+      "step": 1977
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018360087478556228,
+      "loss": 1.4421,
+      "step": 1978
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018358187636079233,
+      "loss": 1.398,
+      "step": 1979
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001835628679216697,
+      "loss": 1.1795,
+      "step": 1980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018354384947047182,
+      "loss": 1.543,
+      "step": 1981
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001835248210094774,
+      "loss": 1.3329,
+      "step": 1982
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018350578254096642,
+      "loss": 1.3951,
+      "step": 1983
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018348673406721986,
+      "loss": 1.2858,
+      "step": 1984
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018346767559052006,
+      "loss": 1.3539,
+      "step": 1985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018344860711315053,
+      "loss": 1.4907,
+      "step": 1986
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018342952863739594,
+      "loss": 1.4074,
+      "step": 1987
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018341044016554215,
+      "loss": 1.2266,
+      "step": 1988
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018339134169987627,
+      "loss": 1.1852,
+      "step": 1989
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018337223324268656,
+      "loss": 1.6029,
+      "step": 1990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001833531147962625,
+      "loss": 1.3563,
+      "step": 1991
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001833339863628948,
+      "loss": 1.2285,
+      "step": 1992
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018331484794487527,
+      "loss": 1.1682,
+      "step": 1993
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018329569954449705,
+      "loss": 0.9792,
+      "step": 1994
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018327654116405436,
+      "loss": 0.8626,
+      "step": 1995
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018325737280584266,
+      "loss": 1.2803,
+      "step": 1996
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018323819447215863,
+      "loss": 1.5574,
+      "step": 1997
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018321900616530008,
+      "loss": 1.5528,
+      "step": 1998
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018319980788756606,
+      "loss": 1.3288,
+      "step": 1999
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018318059964125683,
+      "loss": 1.2006,
+      "step": 2000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018316138142867382,
+      "loss": 1.2191,
+      "step": 2001
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018314215325211967,
+      "loss": 1.2164,
+      "step": 2002
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018312291511389818,
+      "loss": 1.3669,
+      "step": 2003
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018310366701631436,
+      "loss": 1.3682,
+      "step": 2004
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018308440896167445,
+      "loss": 1.0005,
+      "step": 2005
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001830651409522858,
+      "loss": 1.2212,
+      "step": 2006
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001830458629904571,
+      "loss": 1.5365,
+      "step": 2007
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018302657507849804,
+      "loss": 1.2826,
+      "step": 2008
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001830072772187196,
+      "loss": 1.3857,
+      "step": 2009
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000182987969413434,
+      "loss": 1.5468,
+      "step": 2010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001829686516649546,
+      "loss": 1.2372,
+      "step": 2011
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018294932397559593,
+      "loss": 1.4617,
+      "step": 2012
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018292998634767375,
+      "loss": 1.0746,
+      "step": 2013
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000182910638783505,
+      "loss": 1.2666,
+      "step": 2014
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018289128128540778,
+      "loss": 1.2265,
+      "step": 2015
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018287191385570143,
+      "loss": 1.4661,
+      "step": 2016
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018285253649670647,
+      "loss": 1.1818,
+      "step": 2017
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018283314921074454,
+      "loss": 1.4045,
+      "step": 2018
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018281375200013858,
+      "loss": 1.4783,
+      "step": 2019
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018279434486721263,
+      "loss": 1.3622,
+      "step": 2020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018277492781429197,
+      "loss": 1.1517,
+      "step": 2021
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018275550084370307,
+      "loss": 1.4663,
+      "step": 2022
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001827360639577735,
+      "loss": 1.5573,
+      "step": 2023
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018271661715883218,
+      "loss": 1.5934,
+      "step": 2024
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001826971604492091,
+      "loss": 1.3964,
+      "step": 2025
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018267769383123543,
+      "loss": 1.4025,
+      "step": 2026
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018265821730724358,
+      "loss": 1.2638,
+      "step": 2027
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001826387308795671,
+      "loss": 1.2747,
+      "step": 2028
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001826192345505408,
+      "loss": 1.3574,
+      "step": 2029
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001825997283225006,
+      "loss": 1.273,
+      "step": 2030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018258021219778365,
+      "loss": 1.7243,
+      "step": 2031
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018256068617872827,
+      "loss": 1.3322,
+      "step": 2032
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018254115026767397,
+      "loss": 1.407,
+      "step": 2033
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018252160446696145,
+      "loss": 1.2122,
+      "step": 2034
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018250204877893258,
+      "loss": 1.5041,
+      "step": 2035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018248248320593038,
+      "loss": 1.3434,
+      "step": 2036
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018246290775029918,
+      "loss": 1.6837,
+      "step": 2037
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018244332241438435,
+      "loss": 1.2047,
+      "step": 2038
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001824237272005325,
+      "loss": 1.425,
+      "step": 2039
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018240412211109147,
+      "loss": 1.5439,
+      "step": 2040
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018238450714841022,
+      "loss": 1.486,
+      "step": 2041
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018236488231483894,
+      "loss": 1.4633,
+      "step": 2042
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018234524761272892,
+      "loss": 1.525,
+      "step": 2043
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018232560304443272,
+      "loss": 1.1108,
+      "step": 2044
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001823059486123041,
+      "loss": 1.2797,
+      "step": 2045
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018228628431869788,
+      "loss": 1.2908,
+      "step": 2046
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018226661016597015,
+      "loss": 1.3217,
+      "step": 2047
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018224692615647817,
+      "loss": 1.4458,
+      "step": 2048
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018222723229258044,
+      "loss": 1.1725,
+      "step": 2049
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018220752857663647,
+      "loss": 0.6249,
+      "step": 2050
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018218781501100714,
+      "loss": 1.5358,
+      "step": 2051
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001821680915980544,
+      "loss": 1.3139,
+      "step": 2052
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018214835834014142,
+      "loss": 1.3083,
+      "step": 2053
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001821286152396325,
+      "loss": 1.6204,
+      "step": 2054
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001821088622988932,
+      "loss": 1.2099,
+      "step": 2055
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018208909952029017,
+      "loss": 1.7242,
+      "step": 2056
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018206932690619136,
+      "loss": 1.1707,
+      "step": 2057
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018204954445896578,
+      "loss": 1.1355,
+      "step": 2058
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018202975218098362,
+      "loss": 1.2229,
+      "step": 2059
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018200995007461638,
+      "loss": 1.0901,
+      "step": 2060
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018199013814223657,
+      "loss": 1.3383,
+      "step": 2061
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000181970316386218,
+      "loss": 1.3548,
+      "step": 2062
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001819504848089356,
+      "loss": 1.6916,
+      "step": 2063
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018193064341276544,
+      "loss": 1.3576,
+      "step": 2064
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018191079220008488,
+      "loss": 1.4065,
+      "step": 2065
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001818909311732724,
+      "loss": 1.237,
+      "step": 2066
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018187106033470756,
+      "loss": 1.117,
+      "step": 2067
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018185117968677126,
+      "loss": 1.6751,
+      "step": 2068
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018183128923184548,
+      "loss": 1.1849,
+      "step": 2069
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018181138897231338,
+      "loss": 1.4848,
+      "step": 2070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001817914789105593,
+      "loss": 1.3906,
+      "step": 2071
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001817715590489688,
+      "loss": 1.0783,
+      "step": 2072
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001817516293899285,
+      "loss": 1.437,
+      "step": 2073
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018173168993582638,
+      "loss": 1.4869,
+      "step": 2074
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001817117406890514,
+      "loss": 1.3783,
+      "step": 2075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018169178165199377,
+      "loss": 1.6,
+      "step": 2076
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018167181282704499,
+      "loss": 1.1812,
+      "step": 2077
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018165183421659745,
+      "loss": 1.2758,
+      "step": 2078
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018163184582304504,
+      "loss": 1.0608,
+      "step": 2079
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018161184764878258,
+      "loss": 1.5246,
+      "step": 2080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001815918396962062,
+      "loss": 1.7958,
+      "step": 2081
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018157182196771308,
+      "loss": 1.1207,
+      "step": 2082
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018155179446570173,
+      "loss": 1.225,
+      "step": 2083
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018153175719257166,
+      "loss": 1.3778,
+      "step": 2084
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018151171015072368,
+      "loss": 1.219,
+      "step": 2085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018149165334255973,
+      "loss": 1.2545,
+      "step": 2086
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001814715867704829,
+      "loss": 1.4499,
+      "step": 2087
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018145151043689746,
+      "loss": 1.4283,
+      "step": 2088
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018143142434420885,
+      "loss": 1.2091,
+      "step": 2089
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018141132849482368,
+      "loss": 1.4834,
+      "step": 2090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018139122289114978,
+      "loss": 1.7116,
+      "step": 2091
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018137110753559604,
+      "loss": 1.6166,
+      "step": 2092
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018135098243057262,
+      "loss": 1.2883,
+      "step": 2093
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018133084757849076,
+      "loss": 1.0385,
+      "step": 2094
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018131070298176295,
+      "loss": 1.0512,
+      "step": 2095
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018129054864280284,
+      "loss": 1.2537,
+      "step": 2096
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018127038456402513,
+      "loss": 1.2486,
+      "step": 2097
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001812502107478459,
+      "loss": 1.2741,
+      "step": 2098
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018123002719668216,
+      "loss": 1.1998,
+      "step": 2099
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018120983391295225,
+      "loss": 1.5308,
+      "step": 2100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001811896308990756,
+      "loss": 1.3208,
+      "step": 2101
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001811694181574729,
+      "loss": 1.4599,
+      "step": 2102
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018114919569056585,
+      "loss": 1.4621,
+      "step": 2103
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018112896350077747,
+      "loss": 1.2598,
+      "step": 2104
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018110872159053188,
+      "loss": 1.2376,
+      "step": 2105
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001810884699622543,
+      "loss": 1.4472,
+      "step": 2106
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001810682086183712,
+      "loss": 1.2704,
+      "step": 2107
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001810479375613102,
+      "loss": 1.3669,
+      "step": 2108
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001810276567935001,
+      "loss": 1.232,
+      "step": 2109
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018100736631737076,
+      "loss": 1.3798,
+      "step": 2110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001809870661353534,
+      "loss": 1.4487,
+      "step": 2111
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018096675624988014,
+      "loss": 1.1376,
+      "step": 2112
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018094643666338457,
+      "loss": 1.1145,
+      "step": 2113
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018092610737830114,
+      "loss": 1.5858,
+      "step": 2114
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018090576839706568,
+      "loss": 1.4911,
+      "step": 2115
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018088541972211504,
+      "loss": 1.4641,
+      "step": 2116
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018086506135588735,
+      "loss": 1.2525,
+      "step": 2117
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018084469330082183,
+      "loss": 1.4623,
+      "step": 2118
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001808243155593589,
+      "loss": 1.6451,
+      "step": 2119
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018080392813394003,
+      "loss": 1.6568,
+      "step": 2120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018078353102700802,
+      "loss": 1.4058,
+      "step": 2121
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018076312424100674,
+      "loss": 1.1717,
+      "step": 2122
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001807427077783812,
+      "loss": 1.566,
+      "step": 2123
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001807222816415776,
+      "loss": 1.2987,
+      "step": 2124
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018070184583304327,
+      "loss": 1.4279,
+      "step": 2125
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001806814003552268,
+      "loss": 1.579,
+      "step": 2126
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001806609452105778,
+      "loss": 1.139,
+      "step": 2127
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001806404804015471,
+      "loss": 1.5729,
+      "step": 2128
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001806200059305867,
+      "loss": 1.5411,
+      "step": 2129
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001805995218001498,
+      "loss": 1.5269,
+      "step": 2130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018057902801269063,
+      "loss": 1.4584,
+      "step": 2131
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001805585245706647,
+      "loss": 1.0997,
+      "step": 2132
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018053801147652855,
+      "loss": 1.7344,
+      "step": 2133
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018051748873274006,
+      "loss": 1.0304,
+      "step": 2134
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001804969563417581,
+      "loss": 1.2928,
+      "step": 2135
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018047641430604273,
+      "loss": 1.3385,
+      "step": 2136
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018045586262805524,
+      "loss": 1.3702,
+      "step": 2137
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018043530131025802,
+      "loss": 1.2609,
+      "step": 2138
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018041473035511465,
+      "loss": 1.1946,
+      "step": 2139
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018039414976508974,
+      "loss": 1.2258,
+      "step": 2140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018037355954264928,
+      "loss": 1.2178,
+      "step": 2141
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018035295969026015,
+      "loss": 1.5411,
+      "step": 2142
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018033235021039064,
+      "loss": 1.3577,
+      "step": 2143
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018031173110550998,
+      "loss": 1.6105,
+      "step": 2144
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001802911023780887,
+      "loss": 1.4756,
+      "step": 2145
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018027046403059843,
+      "loss": 1.5481,
+      "step": 2146
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001802498160655119,
+      "loss": 1.6068,
+      "step": 2147
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018022915848530307,
+      "loss": 1.4566,
+      "step": 2148
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018020849129244705,
+      "loss": 1.3884,
+      "step": 2149
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001801878144894201,
+      "loss": 1.2419,
+      "step": 2150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018016712807869954,
+      "loss": 1.4474,
+      "step": 2151
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018014643206276394,
+      "loss": 1.5059,
+      "step": 2152
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018012572644409298,
+      "loss": 1.3501,
+      "step": 2153
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001801050112251675,
+      "loss": 1.3524,
+      "step": 2154
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018008428640846954,
+      "loss": 1.2846,
+      "step": 2155
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018006355199648216,
+      "loss": 1.1729,
+      "step": 2156
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018004280799168974,
+      "loss": 1.295,
+      "step": 2157
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018002205439657763,
+      "loss": 1.3113,
+      "step": 2158
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001800012912136325,
+      "loss": 1.2784,
+      "step": 2159
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017998051844534204,
+      "loss": 1.3711,
+      "step": 2160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017995973609419516,
+      "loss": 1.4393,
+      "step": 2161
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017993894416268189,
+      "loss": 1.3938,
+      "step": 2162
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001799181426532934,
+      "loss": 1.4189,
+      "step": 2163
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000179897331568522,
+      "loss": 1.5577,
+      "step": 2164
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017987651091086124,
+      "loss": 1.5492,
+      "step": 2165
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001798556806828057,
+      "loss": 1.4056,
+      "step": 2166
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017983484088685112,
+      "loss": 1.5788,
+      "step": 2167
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017981399152549446,
+      "loss": 1.4745,
+      "step": 2168
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001797931326012338,
+      "loss": 1.6799,
+      "step": 2169
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001797722641165683,
+      "loss": 1.6781,
+      "step": 2170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017975138607399834,
+      "loss": 1.2247,
+      "step": 2171
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017973049847602542,
+      "loss": 1.4575,
+      "step": 2172
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017970960132515217,
+      "loss": 1.4592,
+      "step": 2173
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001796886946238824,
+      "loss": 1.377,
+      "step": 2174
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017966777837472107,
+      "loss": 1.2857,
+      "step": 2175
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017964685258017418,
+      "loss": 1.6539,
+      "step": 2176
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017962591724274903,
+      "loss": 1.2514,
+      "step": 2177
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017960497236495395,
+      "loss": 1.3737,
+      "step": 2178
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017958401794929843,
+      "loss": 1.3445,
+      "step": 2179
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017956305399829313,
+      "loss": 1.7722,
+      "step": 2180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017954208051444988,
+      "loss": 1.4729,
+      "step": 2181
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001795210975002816,
+      "loss": 1.4382,
+      "step": 2182
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017950010495830233,
+      "loss": 1.2168,
+      "step": 2183
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017947910289102734,
+      "loss": 1.2022,
+      "step": 2184
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017945809130097296,
+      "loss": 1.3439,
+      "step": 2185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017943707019065666,
+      "loss": 1.4425,
+      "step": 2186
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017941603956259715,
+      "loss": 1.1627,
+      "step": 2187
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001793949994193142,
+      "loss": 1.4076,
+      "step": 2188
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001793739497633287,
+      "loss": 1.6988,
+      "step": 2189
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017935289059716272,
+      "loss": 1.2519,
+      "step": 2190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017933182192333944,
+      "loss": 1.3335,
+      "step": 2191
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001793107437443833,
+      "loss": 1.1948,
+      "step": 2192
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017928965606281961,
+      "loss": 1.2505,
+      "step": 2193
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017926855888117515,
+      "loss": 1.0113,
+      "step": 2194
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017924745220197763,
+      "loss": 1.2227,
+      "step": 2195
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017922633602775588,
+      "loss": 1.3555,
+      "step": 2196
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017920521036104002,
+      "loss": 1.7118,
+      "step": 2197
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017918407520436114,
+      "loss": 1.2011,
+      "step": 2198
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001791629305602516,
+      "loss": 1.6622,
+      "step": 2199
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001791417764312448,
+      "loss": 1.3809,
+      "step": 2200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001791206128198754,
+      "loss": 1.5308,
+      "step": 2201
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000179099439728679,
+      "loss": 1.3431,
+      "step": 2202
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017907825716019255,
+      "loss": 1.2302,
+      "step": 2203
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017905706511695403,
+      "loss": 0.9714,
+      "step": 2204
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017903586360150248,
+      "loss": 1.213,
+      "step": 2205
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017901465261637825,
+      "loss": 1.0972,
+      "step": 2206
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017899343216412264,
+      "loss": 1.4657,
+      "step": 2207
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017897220224727826,
+      "loss": 1.0624,
+      "step": 2208
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017895096286838874,
+      "loss": 1.2802,
+      "step": 2209
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017892971402999887,
+      "loss": 1.7141,
+      "step": 2210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017890845573465458,
+      "loss": 1.4408,
+      "step": 2211
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017888718798490292,
+      "loss": 1.3291,
+      "step": 2212
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017886591078329208,
+      "loss": 1.247,
+      "step": 2213
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017884462413237143,
+      "loss": 1.2834,
+      "step": 2214
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017882332803469136,
+      "loss": 1.4317,
+      "step": 2215
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017880202249280353,
+      "loss": 1.6,
+      "step": 2216
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001787807075092606,
+      "loss": 0.9082,
+      "step": 2217
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017875938308661643,
+      "loss": 1.6351,
+      "step": 2218
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017873804922742605,
+      "loss": 1.2006,
+      "step": 2219
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017871670593424554,
+      "loss": 1.5793,
+      "step": 2220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017869535320963212,
+      "loss": 1.1349,
+      "step": 2221
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017867399105614423,
+      "loss": 1.285,
+      "step": 2222
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017865261947634133,
+      "loss": 1.7045,
+      "step": 2223
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017863123847278406,
+      "loss": 1.4675,
+      "step": 2224
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017860984804803415,
+      "loss": 1.4418,
+      "step": 2225
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017858844820465458,
+      "loss": 1.4851,
+      "step": 2226
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017856703894520925,
+      "loss": 1.5454,
+      "step": 2227
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001785456202722634,
+      "loss": 1.339,
+      "step": 2228
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001785241921883833,
+      "loss": 1.3066,
+      "step": 2229
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001785027546961363,
+      "loss": 1.4572,
+      "step": 2230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017848130779809097,
+      "loss": 1.3032,
+      "step": 2231
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017845985149681694,
+      "loss": 1.4924,
+      "step": 2232
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017843838579488503,
+      "loss": 1.2954,
+      "step": 2233
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017841691069486713,
+      "loss": 1.4315,
+      "step": 2234
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017839542619933627,
+      "loss": 1.3794,
+      "step": 2235
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017837393231086662,
+      "loss": 1.2918,
+      "step": 2236
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001783524290320335,
+      "loss": 1.0839,
+      "step": 2237
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017833091636541325,
+      "loss": 1.4204,
+      "step": 2238
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001783093943135835,
+      "loss": 1.5336,
+      "step": 2239
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017828786287912282,
+      "loss": 1.2993,
+      "step": 2240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017826632206461107,
+      "loss": 1.3216,
+      "step": 2241
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001782447718726291,
+      "loss": 1.3549,
+      "step": 2242
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000178223212305759,
+      "loss": 1.3764,
+      "step": 2243
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001782016433665839,
+      "loss": 1.2543,
+      "step": 2244
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017818006505768806,
+      "loss": 1.1544,
+      "step": 2245
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001781584773816569,
+      "loss": 1.6641,
+      "step": 2246
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017813688034107696,
+      "loss": 1.2786,
+      "step": 2247
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017811527393853585,
+      "loss": 1.3679,
+      "step": 2248
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017809365817662238,
+      "loss": 1.312,
+      "step": 2249
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017807203305792642,
+      "loss": 1.5667,
+      "step": 2250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000178050398585039,
+      "loss": 1.4705,
+      "step": 2251
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017802875476055218,
+      "loss": 1.3378,
+      "step": 2252
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017800710158705934,
+      "loss": 1.2458,
+      "step": 2253
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017798543906715473,
+      "loss": 1.1,
+      "step": 2254
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001779637672034339,
+      "loss": 1.54,
+      "step": 2255
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017794208599849348,
+      "loss": 1.0634,
+      "step": 2256
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017792039545493115,
+      "loss": 1.306,
+      "step": 2257
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017789869557534579,
+      "loss": 1.3272,
+      "step": 2258
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017787698636233736,
+      "loss": 1.5522,
+      "step": 2259
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017785526781850694,
+      "loss": 1.1129,
+      "step": 2260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017783353994645678,
+      "loss": 1.0387,
+      "step": 2261
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017781180274879015,
+      "loss": 1.5343,
+      "step": 2262
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017779005622811149,
+      "loss": 1.2888,
+      "step": 2263
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001777683003870264,
+      "loss": 0.9308,
+      "step": 2264
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001777465352281415,
+      "loss": 1.3291,
+      "step": 2265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017772476075406462,
+      "loss": 1.108,
+      "step": 2266
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017770297696740467,
+      "loss": 1.6689,
+      "step": 2267
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017768118387077162,
+      "loss": 1.1487,
+      "step": 2268
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017765938146677667,
+      "loss": 1.302,
+      "step": 2269
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017763756975803206,
+      "loss": 1.2088,
+      "step": 2270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001776157487471511,
+      "loss": 1.2051,
+      "step": 2271
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017759391843674832,
+      "loss": 1.2274,
+      "step": 2272
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017757207882943934,
+      "loss": 1.0455,
+      "step": 2273
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017755022992784082,
+      "loss": 1.2819,
+      "step": 2274
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001775283717345706,
+      "loss": 1.0111,
+      "step": 2275
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017750650425224764,
+      "loss": 1.2031,
+      "step": 2276
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017748462748349195,
+      "loss": 1.1459,
+      "step": 2277
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017746274143092471,
+      "loss": 1.2349,
+      "step": 2278
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017744084609716825,
+      "loss": 1.2742,
+      "step": 2279
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017741894148484586,
+      "loss": 1.1312,
+      "step": 2280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017739702759658207,
+      "loss": 1.6464,
+      "step": 2281
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017737510443500254,
+      "loss": 1.4376,
+      "step": 2282
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017735317200273394,
+      "loss": 1.272,
+      "step": 2283
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001773312303024041,
+      "loss": 1.188,
+      "step": 2284
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017730927933664199,
+      "loss": 1.5263,
+      "step": 2285
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017728731910807767,
+      "loss": 1.1417,
+      "step": 2286
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017726534961934226,
+      "loss": 1.3893,
+      "step": 2287
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017724337087306804,
+      "loss": 1.2499,
+      "step": 2288
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017722138287188847,
+      "loss": 1.0373,
+      "step": 2289
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001771993856184379,
+      "loss": 1.3986,
+      "step": 2290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017717737911535203,
+      "loss": 1.0321,
+      "step": 2291
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017715536336526756,
+      "loss": 1.4304,
+      "step": 2292
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017713333837082225,
+      "loss": 1.4511,
+      "step": 2293
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017711130413465507,
+      "loss": 1.1718,
+      "step": 2294
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017708926065940604,
+      "loss": 1.285,
+      "step": 2295
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001770672079477163,
+      "loss": 1.4091,
+      "step": 2296
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017704514600222807,
+      "loss": 1.1121,
+      "step": 2297
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017702307482558472,
+      "loss": 1.3474,
+      "step": 2298
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017700099442043073,
+      "loss": 1.7208,
+      "step": 2299
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001769789047894116,
+      "loss": 1.7352,
+      "step": 2300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017695680593517404,
+      "loss": 1.2522,
+      "step": 2301
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017693469786036583,
+      "loss": 1.4228,
+      "step": 2302
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001769125805676358,
+      "loss": 1.466,
+      "step": 2303
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017689045405963394,
+      "loss": 1.5474,
+      "step": 2304
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017686831833901138,
+      "loss": 1.484,
+      "step": 2305
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017684617340842028,
+      "loss": 1.2011,
+      "step": 2306
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001768240192705139,
+      "loss": 1.886,
+      "step": 2307
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017680185592794671,
+      "loss": 1.0866,
+      "step": 2308
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017677968338337415,
+      "loss": 1.1425,
+      "step": 2309
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017675750163945284,
+      "loss": 1.4211,
+      "step": 2310
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017673531069884048,
+      "loss": 1.4278,
+      "step": 2311
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017671311056419587,
+      "loss": 1.3569,
+      "step": 2312
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017669090123817897,
+      "loss": 1.1498,
+      "step": 2313
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001766686827234507,
+      "loss": 1.6631,
+      "step": 2314
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001766464550226732,
+      "loss": 1.1888,
+      "step": 2315
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017662421813850968,
+      "loss": 1.3208,
+      "step": 2316
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017660197207362445,
+      "loss": 1.353,
+      "step": 2317
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017657971683068295,
+      "loss": 1.4281,
+      "step": 2318
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017655745241235168,
+      "loss": 1.2034,
+      "step": 2319
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017653517882129823,
+      "loss": 1.2869,
+      "step": 2320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001765128960601913,
+      "loss": 1.6574,
+      "step": 2321
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001764906041317007,
+      "loss": 1.2898,
+      "step": 2322
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017646830303849736,
+      "loss": 1.4074,
+      "step": 2323
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017644599278325327,
+      "loss": 1.2434,
+      "step": 2324
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017642367336864151,
+      "loss": 1.2128,
+      "step": 2325
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017640134479733632,
+      "loss": 1.1396,
+      "step": 2326
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017637900707201295,
+      "loss": 1.4818,
+      "step": 2327
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017635666019534784,
+      "loss": 1.5855,
+      "step": 2328
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017633430417001843,
+      "loss": 1.0987,
+      "step": 2329
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017631193899870335,
+      "loss": 1.4105,
+      "step": 2330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017628956468408223,
+      "loss": 1.8133,
+      "step": 2331
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017626718122883586,
+      "loss": 1.2426,
+      "step": 2332
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017624478863564614,
+      "loss": 1.4766,
+      "step": 2333
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017622238690719603,
+      "loss": 1.5103,
+      "step": 2334
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017619997604616957,
+      "loss": 1.2666,
+      "step": 2335
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017617755605525194,
+      "loss": 1.2048,
+      "step": 2336
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017615512693712937,
+      "loss": 1.4314,
+      "step": 2337
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017613268869448917,
+      "loss": 1.4864,
+      "step": 2338
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017611024133001984,
+      "loss": 1.553,
+      "step": 2339
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017608778484641085,
+      "loss": 1.6058,
+      "step": 2340
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001760653192463529,
+      "loss": 1.4702,
+      "step": 2341
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017604284453253763,
+      "loss": 1.5864,
+      "step": 2342
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017602036070765787,
+      "loss": 1.388,
+      "step": 2343
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001759978677744075,
+      "loss": 1.1687,
+      "step": 2344
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001759753657354815,
+      "loss": 1.4157,
+      "step": 2345
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017595285459357603,
+      "loss": 1.2689,
+      "step": 2346
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017593033435138816,
+      "loss": 1.246,
+      "step": 2347
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001759078050116162,
+      "loss": 0.9861,
+      "step": 2348
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001758852665769595,
+      "loss": 1.4461,
+      "step": 2349
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001758627190501185,
+      "loss": 1.3248,
+      "step": 2350
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017584016243379474,
+      "loss": 1.4423,
+      "step": 2351
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017581759673069078,
+      "loss": 1.5138,
+      "step": 2352
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001757950219435104,
+      "loss": 1.4048,
+      "step": 2353
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017577243807495832,
+      "loss": 1.3341,
+      "step": 2354
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001757498451277405,
+      "loss": 1.3951,
+      "step": 2355
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017572724310456388,
+      "loss": 1.2723,
+      "step": 2356
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001757046320081365,
+      "loss": 1.2359,
+      "step": 2357
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001756820118411675,
+      "loss": 1.2943,
+      "step": 2358
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017565938260636715,
+      "loss": 1.4868,
+      "step": 2359
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017563674430644678,
+      "loss": 1.2672,
+      "step": 2360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017561409694411876,
+      "loss": 1.3654,
+      "step": 2361
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017559144052209656,
+      "loss": 1.1961,
+      "step": 2362
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001755687750430948,
+      "loss": 0.9099,
+      "step": 2363
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017554610050982917,
+      "loss": 1.5707,
+      "step": 2364
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017552341692501634,
+      "loss": 1.4018,
+      "step": 2365
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017550072429137417,
+      "loss": 1.3437,
+      "step": 2366
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017547802261162158,
+      "loss": 1.4804,
+      "step": 2367
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017545531188847862,
+      "loss": 1.0676,
+      "step": 2368
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017543259212466627,
+      "loss": 1.1873,
+      "step": 2369
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017540986332290675,
+      "loss": 1.1675,
+      "step": 2370
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017538712548592332,
+      "loss": 1.3225,
+      "step": 2371
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001753643786164403,
+      "loss": 1.5971,
+      "step": 2372
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001753416227171831,
+      "loss": 1.3436,
+      "step": 2373
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017531885779087822,
+      "loss": 1.3972,
+      "step": 2374
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017529608384025323,
+      "loss": 1.565,
+      "step": 2375
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017527330086803682,
+      "loss": 1.3348,
+      "step": 2376
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017525050887695867,
+      "loss": 1.1689,
+      "step": 2377
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017522770786974963,
+      "loss": 1.4376,
+      "step": 2378
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001752048978491416,
+      "loss": 1.2179,
+      "step": 2379
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017518207881786756,
+      "loss": 1.0791,
+      "step": 2380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001751592507786616,
+      "loss": 1.2336,
+      "step": 2381
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017513641373425882,
+      "loss": 1.7763,
+      "step": 2382
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017511356768739545,
+      "loss": 1.2025,
+      "step": 2383
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017509071264080873,
+      "loss": 1.4324,
+      "step": 2384
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017506784859723715,
+      "loss": 1.4498,
+      "step": 2385
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001750449755594201,
+      "loss": 1.4159,
+      "step": 2386
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017502209353009807,
+      "loss": 1.5508,
+      "step": 2387
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017499920251201276,
+      "loss": 1.3813,
+      "step": 2388
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001749763025079068,
+      "loss": 1.5245,
+      "step": 2389
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001749533935205239,
+      "loss": 1.4309,
+      "step": 2390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017493047555260902,
+      "loss": 1.4262,
+      "step": 2391
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017490754860690797,
+      "loss": 1.633,
+      "step": 2392
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001748846126861678,
+      "loss": 1.0907,
+      "step": 2393
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001748616677931365,
+      "loss": 1.4977,
+      "step": 2394
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001748387139305633,
+      "loss": 1.5004,
+      "step": 2395
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001748157511011984,
+      "loss": 1.3405,
+      "step": 2396
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017479277930779302,
+      "loss": 1.3603,
+      "step": 2397
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017476979855309956,
+      "loss": 1.6924,
+      "step": 2398
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001747468088398715,
+      "loss": 1.3358,
+      "step": 2399
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001747238101708633,
+      "loss": 1.6709,
+      "step": 2400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017470080254883055,
+      "loss": 1.4765,
+      "step": 2401
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001746777859765299,
+      "loss": 1.4834,
+      "step": 2402
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001746547604567191,
+      "loss": 1.0127,
+      "step": 2403
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017463172599215695,
+      "loss": 1.0969,
+      "step": 2404
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001746086825856033,
+      "loss": 1.2568,
+      "step": 2405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001745856302398191,
+      "loss": 1.269,
+      "step": 2406
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017456256895756638,
+      "loss": 1.0835,
+      "step": 2407
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001745394987416082,
+      "loss": 1.0207,
+      "step": 2408
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017451641959470872,
+      "loss": 1.2395,
+      "step": 2409
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017449333151963318,
+      "loss": 1.4221,
+      "step": 2410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001744702345191479,
+      "loss": 1.0966,
+      "step": 2411
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017444712859602016,
+      "loss": 1.0961,
+      "step": 2412
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017442401375301845,
+      "loss": 1.6756,
+      "step": 2413
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001744008899929123,
+      "loss": 1.296,
+      "step": 2414
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017437775731847226,
+      "loss": 1.3339,
+      "step": 2415
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001743546157324699,
+      "loss": 1.3944,
+      "step": 2416
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017433146523767803,
+      "loss": 0.9173,
+      "step": 2417
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017430830583687037,
+      "loss": 1.5919,
+      "step": 2418
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017428513753282174,
+      "loss": 1.2829,
+      "step": 2419
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017426196032830813,
+      "loss": 1.7934,
+      "step": 2420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001742387742261064,
+      "loss": 1.5101,
+      "step": 2421
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001742155792289947,
+      "loss": 1.242,
+      "step": 2422
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017419237533975205,
+      "loss": 1.3356,
+      "step": 2423
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017416916256115866,
+      "loss": 1.0978,
+      "step": 2424
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017414594089599578,
+      "loss": 1.2305,
+      "step": 2425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001741227103470457,
+      "loss": 1.4881,
+      "step": 2426
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017409947091709177,
+      "loss": 1.4058,
+      "step": 2427
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017407622260891845,
+      "loss": 1.2883,
+      "step": 2428
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017405296542531118,
+      "loss": 1.1565,
+      "step": 2429
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001740296993690566,
+      "loss": 1.2195,
+      "step": 2430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001740064244429422,
+      "loss": 1.2457,
+      "step": 2431
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001739831406497568,
+      "loss": 1.2116,
+      "step": 2432
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017395984799229005,
+      "loss": 1.1405,
+      "step": 2433
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001739365464733328,
+      "loss": 1.4261,
+      "step": 2434
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017391323609567696,
+      "loss": 1.7817,
+      "step": 2435
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017388991686211538,
+      "loss": 1.2515,
+      "step": 2436
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001738665887754421,
+      "loss": 0.7975,
+      "step": 2437
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001738432518384521,
+      "loss": 1.2151,
+      "step": 2438
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017381990605394161,
+      "loss": 1.2893,
+      "step": 2439
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017379655142470772,
+      "loss": 1.6532,
+      "step": 2440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001737731879535487,
+      "loss": 1.2161,
+      "step": 2441
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001737498156432638,
+      "loss": 1.3367,
+      "step": 2442
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001737264344966534,
+      "loss": 1.3203,
+      "step": 2443
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001737030445165189,
+      "loss": 1.5652,
+      "step": 2444
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017367964570566278,
+      "loss": 1.4562,
+      "step": 2445
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017365623806688856,
+      "loss": 1.4264,
+      "step": 2446
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017363282160300084,
+      "loss": 1.216,
+      "step": 2447
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017360939631680522,
+      "loss": 1.2144,
+      "step": 2448
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001735859622111084,
+      "loss": 1.5922,
+      "step": 2449
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017356251928871822,
+      "loss": 1.1285,
+      "step": 2450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017353906755244335,
+      "loss": 1.4289,
+      "step": 2451
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017351560700509377,
+      "loss": 1.2289,
+      "step": 2452
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017349213764948038,
+      "loss": 1.2028,
+      "step": 2453
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017346865948841513,
+      "loss": 1.2985,
+      "step": 2454
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017344517252471103,
+      "loss": 1.3921,
+      "step": 2455
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017342167676118227,
+      "loss": 1.3414,
+      "step": 2456
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001733981722006439,
+      "loss": 1.1781,
+      "step": 2457
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017337465884591216,
+      "loss": 1.1749,
+      "step": 2458
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017335113669980423,
+      "loss": 1.5299,
+      "step": 2459
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017332760576513852,
+      "loss": 1.5089,
+      "step": 2460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001733040660447343,
+      "loss": 1.0073,
+      "step": 2461
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000173280517541412,
+      "loss": 1.3383,
+      "step": 2462
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017325696025799315,
+      "loss": 1.5857,
+      "step": 2463
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017323339419730017,
+      "loss": 1.5086,
+      "step": 2464
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017320981936215664,
+      "loss": 1.4094,
+      "step": 2465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017318623575538725,
+      "loss": 1.3762,
+      "step": 2466
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017316264337981755,
+      "loss": 0.9905,
+      "step": 2467
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017313904223827437,
+      "loss": 1.2132,
+      "step": 2468
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001731154323335854,
+      "loss": 1.1051,
+      "step": 2469
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001730918136685795,
+      "loss": 1.2408,
+      "step": 2470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017306818624608652,
+      "loss": 1.1451,
+      "step": 2471
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017304455006893739,
+      "loss": 1.5032,
+      "step": 2472
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017302090513996406,
+      "loss": 1.3179,
+      "step": 2473
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017299725146199954,
+      "loss": 1.2256,
+      "step": 2474
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017297358903787792,
+      "loss": 1.3175,
+      "step": 2475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001729499178704343,
+      "loss": 1.1329,
+      "step": 2476
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017292623796250482,
+      "loss": 1.4251,
+      "step": 2477
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017290254931692672,
+      "loss": 1.5885,
+      "step": 2478
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017287885193653822,
+      "loss": 1.4205,
+      "step": 2479
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001728551458241786,
+      "loss": 1.3786,
+      "step": 2480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017283143098268826,
+      "loss": 1.3984,
+      "step": 2481
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017280770741490858,
+      "loss": 1.553,
+      "step": 2482
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017278397512368197,
+      "loss": 1.4009,
+      "step": 2483
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001727602341118519,
+      "loss": 1.2177,
+      "step": 2484
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017273648438226295,
+      "loss": 1.424,
+      "step": 2485
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017271272593776064,
+      "loss": 1.3725,
+      "step": 2486
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017268895878119164,
+      "loss": 1.7933,
+      "step": 2487
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017266518291540354,
+      "loss": 1.3979,
+      "step": 2488
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001726413983432451,
+      "loss": 1.2834,
+      "step": 2489
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017261760506756603,
+      "loss": 1.1279,
+      "step": 2490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017259380309121714,
+      "loss": 1.2606,
+      "step": 2491
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017256999241705025,
+      "loss": 1.2747,
+      "step": 2492
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017254617304791826,
+      "loss": 1.415,
+      "step": 2493
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017252234498667504,
+      "loss": 1.5852,
+      "step": 2494
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017249850823617557,
+      "loss": 1.2467,
+      "step": 2495
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017247466279927585,
+      "loss": 1.1538,
+      "step": 2496
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001724508086788329,
+      "loss": 1.1687,
+      "step": 2497
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017242694587770482,
+      "loss": 1.7095,
+      "step": 2498
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001724030743987507,
+      "loss": 1.0873,
+      "step": 2499
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001723791942448308,
+      "loss": 1.4734,
+      "step": 2500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017235530541880614,
+      "loss": 1.288,
+      "step": 2501
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001723314079235391,
+      "loss": 2.0077,
+      "step": 2502
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001723075017618929,
+      "loss": 1.5476,
+      "step": 2503
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017228358693673185,
+      "loss": 1.3488,
+      "step": 2504
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017225966345092134,
+      "loss": 1.1701,
+      "step": 2505
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017223573130732773,
+      "loss": 1.6342,
+      "step": 2506
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017221179050881845,
+      "loss": 1.3728,
+      "step": 2507
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017218784105826199,
+      "loss": 1.2355,
+      "step": 2508
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001721638829585278,
+      "loss": 1.2071,
+      "step": 2509
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001721399162124864,
+      "loss": 1.4262,
+      "step": 2510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017211594082300948,
+      "loss": 1.5969,
+      "step": 2511
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017209195679296956,
+      "loss": 1.4805,
+      "step": 2512
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001720679641252403,
+      "loss": 1.3558,
+      "step": 2513
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017204396282269636,
+      "loss": 1.2311,
+      "step": 2514
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017201995288821346,
+      "loss": 1.3913,
+      "step": 2515
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001719959343246684,
+      "loss": 1.6877,
+      "step": 2516
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017197190713493887,
+      "loss": 1.4567,
+      "step": 2517
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017194787132190377,
+      "loss": 1.1592,
+      "step": 2518
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017192382688844288,
+      "loss": 1.3709,
+      "step": 2519
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017189977383743715,
+      "loss": 1.6195,
+      "step": 2520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017187571217176846,
+      "loss": 1.2764,
+      "step": 2521
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017185164189431973,
+      "loss": 1.344,
+      "step": 2522
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017182756300797495,
+      "loss": 1.3032,
+      "step": 2523
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017180347551561914,
+      "loss": 1.2512,
+      "step": 2524
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017177937942013837,
+      "loss": 1.2211,
+      "step": 2525
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017175527472441964,
+      "loss": 0.8206,
+      "step": 2526
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017173116143135114,
+      "loss": 1.205,
+      "step": 2527
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001717070395438219,
+      "loss": 1.1185,
+      "step": 2528
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001716829090647222,
+      "loss": 1.5961,
+      "step": 2529
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001716587699969431,
+      "loss": 1.2236,
+      "step": 2530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001716346223433769,
+      "loss": 1.0507,
+      "step": 2531
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017161046610691686,
+      "loss": 1.1257,
+      "step": 2532
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017158630129045723,
+      "loss": 1.3359,
+      "step": 2533
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001715621278968933,
+      "loss": 1.3216,
+      "step": 2534
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001715379459291214,
+      "loss": 1.5296,
+      "step": 2535
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017151375539003891,
+      "loss": 1.5267,
+      "step": 2536
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001714895562825442,
+      "loss": 1.1471,
+      "step": 2537
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017146534860953673,
+      "loss": 1.5881,
+      "step": 2538
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001714411323739169,
+      "loss": 1.3022,
+      "step": 2539
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017141690757858618,
+      "loss": 1.3136,
+      "step": 2540
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017139267422644703,
+      "loss": 1.4267,
+      "step": 2541
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000171368432320403,
+      "loss": 1.378,
+      "step": 2542
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001713441818633587,
+      "loss": 1.2232,
+      "step": 2543
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017131992285821955,
+      "loss": 1.3583,
+      "step": 2544
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017129565530789222,
+      "loss": 1.1528,
+      "step": 2545
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017127137921528433,
+      "loss": 0.8463,
+      "step": 2546
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001712470945833045,
+      "loss": 1.3176,
+      "step": 2547
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017122280141486243,
+      "loss": 1.5737,
+      "step": 2548
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017119849971286872,
+      "loss": 1.3546,
+      "step": 2549
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017117418948023514,
+      "loss": 1.322,
+      "step": 2550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017114987071987443,
+      "loss": 1.3119,
+      "step": 2551
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001711255434347003,
+      "loss": 1.2541,
+      "step": 2552
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001711012076276275,
+      "loss": 1.3556,
+      "step": 2553
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001710768633015719,
+      "loss": 1.4489,
+      "step": 2554
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017105251045945022,
+      "loss": 1.3817,
+      "step": 2555
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017102814910418035,
+      "loss": 1.3776,
+      "step": 2556
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017100377923868115,
+      "loss": 1.6236,
+      "step": 2557
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017097940086587246,
+      "loss": 1.0862,
+      "step": 2558
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001709550139886752,
+      "loss": 1.4314,
+      "step": 2559
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017093061861001124,
+      "loss": 1.2201,
+      "step": 2560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017090621473280358,
+      "loss": 1.1484,
+      "step": 2561
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017088180235997607,
+      "loss": 1.4331,
+      "step": 2562
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001708573814944538,
+      "loss": 1.4535,
+      "step": 2563
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017083295213916265,
+      "loss": 1.7898,
+      "step": 2564
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017080851429702965,
+      "loss": 1.0299,
+      "step": 2565
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017078406797098282,
+      "loss": 1.3579,
+      "step": 2566
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017075961316395118,
+      "loss": 1.5127,
+      "step": 2567
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001707351498788648,
+      "loss": 1.3812,
+      "step": 2568
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017071067811865476,
+      "loss": 1.0984,
+      "step": 2569
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017068619788625312,
+      "loss": 1.3494,
+      "step": 2570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017066170918459298,
+      "loss": 1.2019,
+      "step": 2571
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017063721201660844,
+      "loss": 1.352,
+      "step": 2572
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017061270638523462,
+      "loss": 1.509,
+      "step": 2573
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017058819229340768,
+      "loss": 1.2709,
+      "step": 2574
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017056366974406477,
+      "loss": 1.2916,
+      "step": 2575
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017053913874014404,
+      "loss": 1.492,
+      "step": 2576
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017051459928458469,
+      "loss": 1.5427,
+      "step": 2577
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017049005138032692,
+      "loss": 1.5369,
+      "step": 2578
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001704654950303119,
+      "loss": 1.2556,
+      "step": 2579
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017044093023748185,
+      "loss": 1.4917,
+      "step": 2580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017041635700478004,
+      "loss": 1.2764,
+      "step": 2581
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017039177533515067,
+      "loss": 1.0949,
+      "step": 2582
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000170367185231539,
+      "loss": 1.1299,
+      "step": 2583
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001703425866968913,
+      "loss": 1.5248,
+      "step": 2584
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017031797973415485,
+      "loss": 1.1985,
+      "step": 2585
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017029336434627786,
+      "loss": 1.3228,
+      "step": 2586
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001702687405362097,
+      "loss": 1.331,
+      "step": 2587
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017024410830690062,
+      "loss": 1.2842,
+      "step": 2588
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017021946766130198,
+      "loss": 1.4552,
+      "step": 2589
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017019481860236605,
+      "loss": 1.8442,
+      "step": 2590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017017016113304617,
+      "loss": 1.3786,
+      "step": 2591
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017014549525629668,
+      "loss": 1.1947,
+      "step": 2592
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001701208209750729,
+      "loss": 1.1952,
+      "step": 2593
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017009613829233118,
+      "loss": 1.4441,
+      "step": 2594
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001700714472110289,
+      "loss": 1.1805,
+      "step": 2595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017004674773412438,
+      "loss": 1.6384,
+      "step": 2596
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017002203986457698,
+      "loss": 1.1279,
+      "step": 2597
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001699973236053471,
+      "loss": 1.4573,
+      "step": 2598
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016997259895939614,
+      "loss": 1.2867,
+      "step": 2599
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016994786592968643,
+      "loss": 1.2397,
+      "step": 2600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016992312451918136,
+      "loss": 1.2393,
+      "step": 2601
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016989837473084537,
+      "loss": 1.3212,
+      "step": 2602
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001698736165676438,
+      "loss": 1.5043,
+      "step": 2603
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016984885003254308,
+      "loss": 1.149,
+      "step": 2604
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016982407512851054,
+      "loss": 1.562,
+      "step": 2605
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001697992918585147,
+      "loss": 1.5359,
+      "step": 2606
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016977450022552483,
+      "loss": 1.0703,
+      "step": 2607
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016974970023251148,
+      "loss": 1.2201,
+      "step": 2608
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016972489188244594,
+      "loss": 0.9373,
+      "step": 2609
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016970007517830066,
+      "loss": 1.2638,
+      "step": 2610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016967525012304905,
+      "loss": 1.328,
+      "step": 2611
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016965041671966553,
+      "loss": 1.3341,
+      "step": 2612
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016962557497112552,
+      "loss": 1.1974,
+      "step": 2613
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001696007248804054,
+      "loss": 1.8,
+      "step": 2614
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001695758664504826,
+      "loss": 1.4454,
+      "step": 2615
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016955099968433553,
+      "loss": 1.275,
+      "step": 2616
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016952612458494363,
+      "loss": 1.6891,
+      "step": 2617
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016950124115528724,
+      "loss": 1.5009,
+      "step": 2618
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001694763493983478,
+      "loss": 1.5267,
+      "step": 2619
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016945144931710773,
+      "loss": 1.63,
+      "step": 2620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001694265409145504,
+      "loss": 1.4294,
+      "step": 2621
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016940162419366018,
+      "loss": 0.9868,
+      "step": 2622
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016937669915742255,
+      "loss": 1.3502,
+      "step": 2623
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016935176580882386,
+      "loss": 1.1915,
+      "step": 2624
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016932682415085146,
+      "loss": 1.1503,
+      "step": 2625
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001693018741864938,
+      "loss": 1.255,
+      "step": 2626
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016927691591874018,
+      "loss": 1.1578,
+      "step": 2627
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000169251949350581,
+      "loss": 1.4755,
+      "step": 2628
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016922697448500768,
+      "loss": 0.8451,
+      "step": 2629
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016920199132501254,
+      "loss": 1.299,
+      "step": 2630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001691769998735889,
+      "loss": 1.3199,
+      "step": 2631
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016915200013373116,
+      "loss": 1.1379,
+      "step": 2632
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016912699210843462,
+      "loss": 1.6992,
+      "step": 2633
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016910197580069568,
+      "loss": 1.335,
+      "step": 2634
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016907695121351157,
+      "loss": 1.1967,
+      "step": 2635
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016905191834988068,
+      "loss": 1.6514,
+      "step": 2636
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001690268772128023,
+      "loss": 1.3221,
+      "step": 2637
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016900182780527672,
+      "loss": 1.6302,
+      "step": 2638
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016897677013030525,
+      "loss": 1.6058,
+      "step": 2639
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016895170419089017,
+      "loss": 1.5251,
+      "step": 2640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016892662999003474,
+      "loss": 1.1031,
+      "step": 2641
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001689015475307432,
+      "loss": 1.1424,
+      "step": 2642
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016887645681602087,
+      "loss": 1.1951,
+      "step": 2643
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001688513578488739,
+      "loss": 1.2156,
+      "step": 2644
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016882625063230962,
+      "loss": 1.4303,
+      "step": 2645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001688011351693362,
+      "loss": 1.3363,
+      "step": 2646
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016877601146296283,
+      "loss": 1.2062,
+      "step": 2647
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016875087951619973,
+      "loss": 1.5379,
+      "step": 2648
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016872573933205805,
+      "loss": 1.0459,
+      "step": 2649
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016870059091355,
+      "loss": 1.3976,
+      "step": 2650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001686754342636887,
+      "loss": 1.2235,
+      "step": 2651
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001686502693854883,
+      "loss": 1.3093,
+      "step": 2652
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016862509628196395,
+      "loss": 1.125,
+      "step": 2653
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016859991495613174,
+      "loss": 1.3433,
+      "step": 2654
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016857472541100874,
+      "loss": 1.4358,
+      "step": 2655
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001685495276496131,
+      "loss": 1.3322,
+      "step": 2656
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016852432167496385,
+      "loss": 1.322,
+      "step": 2657
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016849910749008103,
+      "loss": 1.4569,
+      "step": 2658
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016847388509798566,
+      "loss": 1.4956,
+      "step": 2659
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016844865450169982,
+      "loss": 1.2005,
+      "step": 2660
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016842341570424648,
+      "loss": 1.2881,
+      "step": 2661
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016839816870864959,
+      "loss": 1.1639,
+      "step": 2662
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016837291351793417,
+      "loss": 1.1992,
+      "step": 2663
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016834765013512613,
+      "loss": 1.1743,
+      "step": 2664
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016832237856325243,
+      "loss": 1.6552,
+      "step": 2665
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016829709880534095,
+      "loss": 1.5212,
+      "step": 2666
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016827181086442056,
+      "loss": 1.244,
+      "step": 2667
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016824651474352123,
+      "loss": 1.1953,
+      "step": 2668
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001682212104456737,
+      "loss": 1.2939,
+      "step": 2669
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016819589797390986,
+      "loss": 1.4618,
+      "step": 2670
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001681705773312625,
+      "loss": 1.2844,
+      "step": 2671
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016814524852076543,
+      "loss": 1.3699,
+      "step": 2672
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001681199115454534,
+      "loss": 1.4523,
+      "step": 2673
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001680945664083622,
+      "loss": 1.089,
+      "step": 2674
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016806921311252846,
+      "loss": 1.4783,
+      "step": 2675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016804385166098996,
+      "loss": 1.0934,
+      "step": 2676
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001680184820567854,
+      "loss": 1.5375,
+      "step": 2677
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016799310430295434,
+      "loss": 1.4579,
+      "step": 2678
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016796771840253748,
+      "loss": 1.4434,
+      "step": 2679
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016794232435857643,
+      "loss": 1.2685,
+      "step": 2680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001679169221741138,
+      "loss": 1.4875,
+      "step": 2681
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016789151185219305,
+      "loss": 1.3998,
+      "step": 2682
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001678660933958588,
+      "loss": 0.9823,
+      "step": 2683
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016784066680815656,
+      "loss": 1.3755,
+      "step": 2684
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016781523209213276,
+      "loss": 1.0569,
+      "step": 2685
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016778978925083488,
+      "loss": 1.208,
+      "step": 2686
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001677643382873114,
+      "loss": 1.2435,
+      "step": 2687
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016773887920461163,
+      "loss": 1.2085,
+      "step": 2688
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016771341200578604,
+      "loss": 1.4521,
+      "step": 2689
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001676879366938859,
+      "loss": 1.4596,
+      "step": 2690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001676624532719636,
+      "loss": 1.467,
+      "step": 2691
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001676369617430724,
+      "loss": 1.2039,
+      "step": 2692
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016761146211026655,
+      "loss": 1.2415,
+      "step": 2693
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001675859543766013,
+      "loss": 1.6762,
+      "step": 2694
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001675604385451329,
+      "loss": 1.5673,
+      "step": 2695
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016753491461891847,
+      "loss": 1.2641,
+      "step": 2696
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016750938260101617,
+      "loss": 1.1777,
+      "step": 2697
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016748384249448507,
+      "loss": 1.3332,
+      "step": 2698
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016745829430238538,
+      "loss": 1.4701,
+      "step": 2699
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016743273802777804,
+      "loss": 0.9333,
+      "step": 2700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001674071736737251,
+      "loss": 1.7623,
+      "step": 2701
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016738160124328956,
+      "loss": 1.1791,
+      "step": 2702
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016735602073953538,
+      "loss": 1.3997,
+      "step": 2703
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001673304321655275,
+      "loss": 1.4681,
+      "step": 2704
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016730483552433175,
+      "loss": 1.1534,
+      "step": 2705
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016727923081901507,
+      "loss": 1.1066,
+      "step": 2706
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016725361805264524,
+      "loss": 1.2157,
+      "step": 2707
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016722799722829106,
+      "loss": 1.5676,
+      "step": 2708
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016720236834902227,
+      "loss": 0.9648,
+      "step": 2709
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001671767314179096,
+      "loss": 1.3015,
+      "step": 2710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016715108643802475,
+      "loss": 1.2984,
+      "step": 2711
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016712543341244036,
+      "loss": 1.6145,
+      "step": 2712
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016709977234423006,
+      "loss": 1.7289,
+      "step": 2713
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016707410323646839,
+      "loss": 1.7677,
+      "step": 2714
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016704842609223093,
+      "loss": 1.2335,
+      "step": 2715
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016702274091459418,
+      "loss": 1.2059,
+      "step": 2716
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016699704770663554,
+      "loss": 1.4051,
+      "step": 2717
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016697134647143355,
+      "loss": 1.219,
+      "step": 2718
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001669456372120675,
+      "loss": 1.3064,
+      "step": 2719
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016691991993161783,
+      "loss": 1.5258,
+      "step": 2720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001668941946331658,
+      "loss": 1.4926,
+      "step": 2721
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016686846131979365,
+      "loss": 1.6239,
+      "step": 2722
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001668427199945847,
+      "loss": 1.3266,
+      "step": 2723
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016681697066062308,
+      "loss": 1.3479,
+      "step": 2724
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016679121332099394,
+      "loss": 1.3531,
+      "step": 2725
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016676544797878347,
+      "loss": 1.2848,
+      "step": 2726
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016673967463707865,
+      "loss": 1.5816,
+      "step": 2727
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016671389329896757,
+      "loss": 1.2251,
+      "step": 2728
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016668810396753915,
+      "loss": 1.0445,
+      "step": 2729
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016666230664588342,
+      "loss": 1.2866,
+      "step": 2730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016663650133709123,
+      "loss": 1.8421,
+      "step": 2731
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016661068804425447,
+      "loss": 1.2682,
+      "step": 2732
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016658486677046592,
+      "loss": 1.2303,
+      "step": 2733
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001665590375188194,
+      "loss": 1.3099,
+      "step": 2734
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001665332002924096,
+      "loss": 1.6171,
+      "step": 2735
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001665073550943322,
+      "loss": 1.5939,
+      "step": 2736
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016648150192768389,
+      "loss": 1.2004,
+      "step": 2737
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001664556407955622,
+      "loss": 1.0938,
+      "step": 2738
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001664297717010657,
+      "loss": 1.3164,
+      "step": 2739
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016640389464729394,
+      "loss": 1.2164,
+      "step": 2740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016637800963734727,
+      "loss": 1.1564,
+      "step": 2741
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016635211667432725,
+      "loss": 1.0463,
+      "step": 2742
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001663262157613361,
+      "loss": 1.2258,
+      "step": 2743
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016630030690147724,
+      "loss": 1.0856,
+      "step": 2744
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016627439009785487,
+      "loss": 1.196,
+      "step": 2745
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016624846535357422,
+      "loss": 1.096,
+      "step": 2746
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016622253267174145,
+      "loss": 1.6238,
+      "step": 2747
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016619659205546373,
+      "loss": 1.5172,
+      "step": 2748
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016617064350784913,
+      "loss": 1.3002,
+      "step": 2749
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001661446870320066,
+      "loss": 1.7079,
+      "step": 2750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016611872263104616,
+      "loss": 1.2863,
+      "step": 2751
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016609275030807875,
+      "loss": 1.2091,
+      "step": 2752
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001660667700662162,
+      "loss": 1.1636,
+      "step": 2753
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016604078190857137,
+      "loss": 1.299,
+      "step": 2754
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000166014785838258,
+      "loss": 1.2938,
+      "step": 2755
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016598878185839083,
+      "loss": 1.4463,
+      "step": 2756
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016596276997208546,
+      "loss": 1.1671,
+      "step": 2757
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016593675018245857,
+      "loss": 1.4595,
+      "step": 2758
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001659107224926277,
+      "loss": 0.9981,
+      "step": 2759
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016588468690571136,
+      "loss": 1.3619,
+      "step": 2760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000165858643424829,
+      "loss": 1.464,
+      "step": 2761
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016583259205310097,
+      "loss": 1.4533,
+      "step": 2762
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001658065327936487,
+      "loss": 1.67,
+      "step": 2763
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001657804656495944,
+      "loss": 1.2865,
+      "step": 2764
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016575439062406134,
+      "loss": 1.2098,
+      "step": 2765
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016572830772017367,
+      "loss": 1.6039,
+      "step": 2766
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016570221694105653,
+      "loss": 1.0151,
+      "step": 2767
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016567611828983595,
+      "loss": 1.4261,
+      "step": 2768
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000165650011769639,
+      "loss": 1.4229,
+      "step": 2769
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016562389738359362,
+      "loss": 1.2897,
+      "step": 2770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016559777513482863,
+      "loss": 1.7523,
+      "step": 2771
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016557164502647393,
+      "loss": 1.6747,
+      "step": 2772
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016554550706166028,
+      "loss": 1.6108,
+      "step": 2773
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016551936124351938,
+      "loss": 1.1793,
+      "step": 2774
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016549320757518394,
+      "loss": 1.379,
+      "step": 2775
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001654670460597875,
+      "loss": 1.3027,
+      "step": 2776
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016544087670046463,
+      "loss": 1.2521,
+      "step": 2777
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016541469950035078,
+      "loss": 1.7219,
+      "step": 2778
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001653885144625824,
+      "loss": 1.3961,
+      "step": 2779
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016536232159029683,
+      "loss": 1.2838,
+      "step": 2780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016533612088663236,
+      "loss": 0.8717,
+      "step": 2781
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001653099123547283,
+      "loss": 1.5491,
+      "step": 2782
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001652836959977247,
+      "loss": 1.6295,
+      "step": 2783
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016525747181876272,
+      "loss": 1.3279,
+      "step": 2784
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016523123982098442,
+      "loss": 1.6695,
+      "step": 2785
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001652050000075328,
+      "loss": 1.5738,
+      "step": 2786
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016517875238155174,
+      "loss": 1.0871,
+      "step": 2787
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016515249694618609,
+      "loss": 1.3203,
+      "step": 2788
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001651262337045817,
+      "loss": 1.5623,
+      "step": 2789
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016509996265988525,
+      "loss": 1.1253,
+      "step": 2790
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016507368381524443,
+      "loss": 1.3648,
+      "step": 2791
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016504739717380782,
+      "loss": 1.3231,
+      "step": 2792
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016502110273872495,
+      "loss": 1.1324,
+      "step": 2793
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016499480051314626,
+      "loss": 1.3486,
+      "step": 2794
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001649684905002232,
+      "loss": 1.4045,
+      "step": 2795
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001649421727031081,
+      "loss": 1.111,
+      "step": 2796
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016491584712495416,
+      "loss": 1.243,
+      "step": 2797
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016488951376891563,
+      "loss": 1.0831,
+      "step": 2798
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016486317263814763,
+      "loss": 1.385,
+      "step": 2799
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016483682373580623,
+      "loss": 1.3753,
+      "step": 2800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016481046706504836,
+      "loss": 1.1906,
+      "step": 2801
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016478410262903202,
+      "loss": 1.2793,
+      "step": 2802
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016475773043091602,
+      "loss": 1.3078,
+      "step": 2803
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016473135047386013,
+      "loss": 1.077,
+      "step": 2804
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001647049627610251,
+      "loss": 1.5811,
+      "step": 2805
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016467856729557255,
+      "loss": 1.3902,
+      "step": 2806
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016465216408066504,
+      "loss": 1.4328,
+      "step": 2807
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001646257531194661,
+      "loss": 1.4183,
+      "step": 2808
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016459933441514014,
+      "loss": 1.2959,
+      "step": 2809
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016457290797085248,
+      "loss": 1.4888,
+      "step": 2810
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016454647378976947,
+      "loss": 1.7334,
+      "step": 2811
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016452003187505823,
+      "loss": 1.4046,
+      "step": 2812
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016449358222988698,
+      "loss": 1.3273,
+      "step": 2813
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016446712485742476,
+      "loss": 1.526,
+      "step": 2814
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001644406597608415,
+      "loss": 1.2289,
+      "step": 2815
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016441418694330818,
+      "loss": 1.2583,
+      "step": 2816
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001643877064079966,
+      "loss": 1.2668,
+      "step": 2817
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016436121815807954,
+      "loss": 1.2108,
+      "step": 2818
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016433472219673067,
+      "loss": 1.3972,
+      "step": 2819
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016430821852712465,
+      "loss": 1.2312,
+      "step": 2820
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016428170715243693,
+      "loss": 0.8569,
+      "step": 2821
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016425518807584402,
+      "loss": 1.3837,
+      "step": 2822
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016422866130052333,
+      "loss": 1.3148,
+      "step": 2823
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016420212682965307,
+      "loss": 1.4956,
+      "step": 2824
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016417558466641255,
+      "loss": 1.4386,
+      "step": 2825
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001641490348139819,
+      "loss": 1.803,
+      "step": 2826
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016412247727554217,
+      "loss": 1.1981,
+      "step": 2827
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016409591205427532,
+      "loss": 1.4951,
+      "step": 2828
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001640693391533643,
+      "loss": 1.186,
+      "step": 2829
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016404275857599297,
+      "loss": 1.0122,
+      "step": 2830
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000164016170325346,
+      "loss": 1.2833,
+      "step": 2831
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001639895744046091,
+      "loss": 1.6444,
+      "step": 2832
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016396297081696886,
+      "loss": 1.3589,
+      "step": 2833
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016393635956561277,
+      "loss": 1.6937,
+      "step": 2834
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016390974065372927,
+      "loss": 1.3464,
+      "step": 2835
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001638831140845077,
+      "loss": 1.524,
+      "step": 2836
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016385647986113833,
+      "loss": 1.3535,
+      "step": 2837
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001638298379868123,
+      "loss": 1.3673,
+      "step": 2838
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001638031884647217,
+      "loss": 1.1151,
+      "step": 2839
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001637765312980596,
+      "loss": 1.7757,
+      "step": 2840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016374986649001988,
+      "loss": 1.452,
+      "step": 2841
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016372319404379736,
+      "loss": 1.3191,
+      "step": 2842
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001636965139625879,
+      "loss": 1.2019,
+      "step": 2843
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016366982624958803,
+      "loss": 1.397,
+      "step": 2844
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001636431309079954,
+      "loss": 1.2047,
+      "step": 2845
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016361642794100856,
+      "loss": 1.6226,
+      "step": 2846
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016358971735182682,
+      "loss": 1.1528,
+      "step": 2847
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016356299914365062,
+      "loss": 1.6516,
+      "step": 2848
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016353627331968106,
+      "loss": 1.2451,
+      "step": 2849
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016350953988312044,
+      "loss": 1.2355,
+      "step": 2850
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016348279883717174,
+      "loss": 1.3647,
+      "step": 2851
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001634560501850389,
+      "loss": 1.3323,
+      "step": 2852
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001634292939299269,
+      "loss": 1.0604,
+      "step": 2853
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016340253007504147,
+      "loss": 1.6059,
+      "step": 2854
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016337575862358938,
+      "loss": 1.207,
+      "step": 2855
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016334897957877818,
+      "loss": 1.4877,
+      "step": 2856
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016332219294381642,
+      "loss": 1.5329,
+      "step": 2857
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016329539872191355,
+      "loss": 1.4526,
+      "step": 2858
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016326859691627994,
+      "loss": 1.3127,
+      "step": 2859
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016324178753012678,
+      "loss": 0.9366,
+      "step": 2860
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001632149705666663,
+      "loss": 1.5273,
+      "step": 2861
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016318814602911153,
+      "loss": 1.1861,
+      "step": 2862
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016316131392067647,
+      "loss": 1.1606,
+      "step": 2863
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016313447424457596,
+      "loss": 1.7474,
+      "step": 2864
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016310762700402583,
+      "loss": 1.2905,
+      "step": 2865
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001630807722022428,
+      "loss": 1.244,
+      "step": 2866
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001630539098424445,
+      "loss": 1.2727,
+      "step": 2867
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016302703992784933,
+      "loss": 1.2604,
+      "step": 2868
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016300016246167677,
+      "loss": 1.4779,
+      "step": 2869
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016297327744714717,
+      "loss": 1.5042,
+      "step": 2870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016294638488748172,
+      "loss": 1.4614,
+      "step": 2871
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016291948478590253,
+      "loss": 1.4235,
+      "step": 2872
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016289257714563268,
+      "loss": 1.1749,
+      "step": 2873
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016286566196989605,
+      "loss": 1.2441,
+      "step": 2874
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016283873926191755,
+      "loss": 1.351,
+      "step": 2875
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016281180902492283,
+      "loss": 1.107,
+      "step": 2876
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016278487126213863,
+      "loss": 1.2901,
+      "step": 2877
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016275792597679245,
+      "loss": 1.2234,
+      "step": 2878
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001627309731721127,
+      "loss": 1.3263,
+      "step": 2879
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016270401285132876,
+      "loss": 1.3925,
+      "step": 2880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016267704501767095,
+      "loss": 1.3341,
+      "step": 2881
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016265006967437027,
+      "loss": 1.308,
+      "step": 2882
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001626230868246589,
+      "loss": 1.4562,
+      "step": 2883
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001625960964717697,
+      "loss": 1.609,
+      "step": 2884
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016256909861893655,
+      "loss": 1.5945,
+      "step": 2885
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016254209326939422,
+      "loss": 1.4508,
+      "step": 2886
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001625150804263783,
+      "loss": 1.6294,
+      "step": 2887
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016248806009312537,
+      "loss": 1.5195,
+      "step": 2888
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016246103227287283,
+      "loss": 1.6038,
+      "step": 2889
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016243399696885908,
+      "loss": 1.1981,
+      "step": 2890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016240695418432327,
+      "loss": 1.2652,
+      "step": 2891
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016237990392250562,
+      "loss": 1.5622,
+      "step": 2892
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016235284618664709,
+      "loss": 1.273,
+      "step": 2893
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016232578097998958,
+      "loss": 1.3806,
+      "step": 2894
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016229870830577597,
+      "loss": 1.3226,
+      "step": 2895
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016227162816724987,
+      "loss": 1.4376,
+      "step": 2896
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016224454056765604,
+      "loss": 1.2165,
+      "step": 2897
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016221744551023982,
+      "loss": 1.1811,
+      "step": 2898
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016219034299824771,
+      "loss": 1.2213,
+      "step": 2899
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016216323303492697,
+      "loss": 1.2606,
+      "step": 2900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001621361156235257,
+      "loss": 1.4511,
+      "step": 2901
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016210899076729308,
+      "loss": 1.3388,
+      "step": 2902
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000162081858469479,
+      "loss": 1.1723,
+      "step": 2903
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016205471873333434,
+      "loss": 1.408,
+      "step": 2904
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016202757156211084,
+      "loss": 1.343,
+      "step": 2905
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016200041695906112,
+      "loss": 1.2606,
+      "step": 2906
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016197325492743874,
+      "loss": 1.5151,
+      "step": 2907
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016194608547049808,
+      "loss": 1.327,
+      "step": 2908
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016191890859149449,
+      "loss": 1.2144,
+      "step": 2909
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001618917242936841,
+      "loss": 0.936,
+      "step": 2910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016186453258032403,
+      "loss": 1.4371,
+      "step": 2911
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016183733345467226,
+      "loss": 1.2765,
+      "step": 2912
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016181012691998765,
+      "loss": 1.4298,
+      "step": 2913
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001617829129795299,
+      "loss": 1.544,
+      "step": 2914
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016175569163655974,
+      "loss": 1.7091,
+      "step": 2915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016172846289433862,
+      "loss": 1.2349,
+      "step": 2916
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016170122675612895,
+      "loss": 1.3689,
+      "step": 2917
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016167398322519403,
+      "loss": 1.073,
+      "step": 2918
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016164673230479806,
+      "loss": 1.0343,
+      "step": 2919
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016161947399820608,
+      "loss": 1.3542,
+      "step": 2920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016159220830868408,
+      "loss": 1.3678,
+      "step": 2921
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016156493523949888,
+      "loss": 1.6533,
+      "step": 2922
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016153765479391815,
+      "loss": 1.2544,
+      "step": 2923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016151036697521057,
+      "loss": 1.2621,
+      "step": 2924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016148307178664556,
+      "loss": 1.2407,
+      "step": 2925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016145576923149357,
+      "loss": 1.8104,
+      "step": 2926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016142845931302574,
+      "loss": 1.3585,
+      "step": 2927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016140114203451434,
+      "loss": 1.1497,
+      "step": 2928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016137381739923226,
+      "loss": 1.2362,
+      "step": 2929
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001613464854104535,
+      "loss": 1.1596,
+      "step": 2930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016131914607145272,
+      "loss": 1.2622,
+      "step": 2931
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001612917993855057,
+      "loss": 1.2156,
+      "step": 2932
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001612644453558889,
+      "loss": 1.397,
+      "step": 2933
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001612370839858798,
+      "loss": 1.1934,
+      "step": 2934
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016120971527875666,
+      "loss": 1.5632,
+      "step": 2935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016118233923779866,
+      "loss": 1.8093,
+      "step": 2936
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016115495586628589,
+      "loss": 1.5499,
+      "step": 2937
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016112756516749923,
+      "loss": 1.5239,
+      "step": 2938
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016110016714472055,
+      "loss": 1.1493,
+      "step": 2939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001610727618012325,
+      "loss": 1.0581,
+      "step": 2940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016104534914031867,
+      "loss": 1.6143,
+      "step": 2941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001610179291652635,
+      "loss": 1.591,
+      "step": 2942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001609905018793523,
+      "loss": 1.6231,
+      "step": 2943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001609630672858713,
+      "loss": 1.2352,
+      "step": 2944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001609356253881075,
+      "loss": 1.122,
+      "step": 2945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016090817618934894,
+      "loss": 1.6479,
+      "step": 2946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001608807196928844,
+      "loss": 1.2847,
+      "step": 2947
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016085325590200355,
+      "loss": 1.3692,
+      "step": 2948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000160825784819997,
+      "loss": 1.2805,
+      "step": 2949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016079830645015616,
+      "loss": 1.2351,
+      "step": 2950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016077082079577338,
+      "loss": 1.5047,
+      "step": 2951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016074332786014187,
+      "loss": 1.2763,
+      "step": 2952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016071582764655564,
+      "loss": 1.3782,
+      "step": 2953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016068832015830967,
+      "loss": 1.4763,
+      "step": 2954
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016066080539869971,
+      "loss": 1.2455,
+      "step": 2955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016063328337102248,
+      "loss": 1.5436,
+      "step": 2956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001606057540785755,
+      "loss": 1.4336,
+      "step": 2957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016057821752465728,
+      "loss": 1.3527,
+      "step": 2958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016055067371256696,
+      "loss": 1.2538,
+      "step": 2959
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016052312264560483,
+      "loss": 1.7241,
+      "step": 2960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016049556432707184,
+      "loss": 1.3632,
+      "step": 2961
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016046799876026993,
+      "loss": 1.5892,
+      "step": 2962
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016044042594850181,
+      "loss": 1.5265,
+      "step": 2963
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016041284589507116,
+      "loss": 0.945,
+      "step": 2964
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016038525860328248,
+      "loss": 1.3254,
+      "step": 2965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016035766407644113,
+      "loss": 1.2896,
+      "step": 2966
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016033006231785336,
+      "loss": 1.325,
+      "step": 2967
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016030245333082622,
+      "loss": 1.6178,
+      "step": 2968
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016027483711866772,
+      "loss": 1.3189,
+      "step": 2969
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001602472136846867,
+      "loss": 1.4914,
+      "step": 2970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016021958303219286,
+      "loss": 1.4905,
+      "step": 2971
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001601919451644967,
+      "loss": 1.3041,
+      "step": 2972
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016016430008490971,
+      "loss": 1.2367,
+      "step": 2973
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001601366477967442,
+      "loss": 1.1867,
+      "step": 2974
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016010898830331327,
+      "loss": 1.4086,
+      "step": 2975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016008132160793097,
+      "loss": 1.1324,
+      "step": 2976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00016005364771391216,
+      "loss": 1.3742,
+      "step": 2977
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00016002596662457262,
+      "loss": 1.8133,
+      "step": 2978
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015999827834322893,
+      "loss": 1.5759,
+      "step": 2979
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001599705828731985,
+      "loss": 1.3361,
+      "step": 2980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001599428802177998,
+      "loss": 1.23,
+      "step": 2981
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015991517038035192,
+      "loss": 1.4024,
+      "step": 2982
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015988745336417492,
+      "loss": 1.0739,
+      "step": 2983
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001598597291725897,
+      "loss": 1.2758,
+      "step": 2984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015983199780891813,
+      "loss": 1.2343,
+      "step": 2985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001598042592764827,
+      "loss": 1.1414,
+      "step": 2986
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015977651357860698,
+      "loss": 1.4467,
+      "step": 2987
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015974876071861535,
+      "loss": 1.4237,
+      "step": 2988
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015972100069983292,
+      "loss": 1.3698,
+      "step": 2989
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001596932335255858,
+      "loss": 1.2429,
+      "step": 2990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015966545919920093,
+      "loss": 1.2149,
+      "step": 2991
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015963767772400606,
+      "loss": 1.2039,
+      "step": 2992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015960988910332984,
+      "loss": 1.4448,
+      "step": 2993
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015958209334050177,
+      "loss": 1.1971,
+      "step": 2994
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015955429043885215,
+      "loss": 1.1754,
+      "step": 2995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001595264804017122,
+      "loss": 1.0919,
+      "step": 2996
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015949866323241403,
+      "loss": 1.2666,
+      "step": 2997
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001594708389342905,
+      "loss": 1.2693,
+      "step": 2998
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015944300751067535,
+      "loss": 1.2349,
+      "step": 2999
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015941516896490328,
+      "loss": 1.2075,
+      "step": 3000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015938732330030968,
+      "loss": 1.4623,
+      "step": 3001
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001593594705202309,
+      "loss": 1.3652,
+      "step": 3002
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015933161062800416,
+      "loss": 1.5739,
+      "step": 3003
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015930374362696745,
+      "loss": 1.3116,
+      "step": 3004
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015927586952045966,
+      "loss": 1.2897,
+      "step": 3005
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015924798831182049,
+      "loss": 1.4177,
+      "step": 3006
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015922010000439058,
+      "loss": 1.2428,
+      "step": 3007
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015919220460151133,
+      "loss": 1.4532,
+      "step": 3008
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015916430210652504,
+      "loss": 1.3994,
+      "step": 3009
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015913639252277485,
+      "loss": 1.3249,
+      "step": 3010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001591084758536047,
+      "loss": 1.354,
+      "step": 3011
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001590805521023595,
+      "loss": 1.4636,
+      "step": 3012
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015905262127238485,
+      "loss": 1.3767,
+      "step": 3013
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001590246833670273,
+      "loss": 1.3304,
+      "step": 3014
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015899673838963427,
+      "loss": 1.2974,
+      "step": 3015
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015896878634355396,
+      "loss": 1.4886,
+      "step": 3016
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001589408272321354,
+      "loss": 1.3189,
+      "step": 3017
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001589128610587286,
+      "loss": 1.4194,
+      "step": 3018
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001588848878266842,
+      "loss": 1.3349,
+      "step": 3019
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015885690753935391,
+      "loss": 1.2363,
+      "step": 3020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015882892020009014,
+      "loss": 0.9849,
+      "step": 3021
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001588009258122462,
+      "loss": 1.2499,
+      "step": 3022
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001587729243791763,
+      "loss": 1.3408,
+      "step": 3023
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001587449159042353,
+      "loss": 1.1403,
+      "step": 3024
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001587169003907791,
+      "loss": 1.1755,
+      "step": 3025
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015868887784216437,
+      "loss": 1.5679,
+      "step": 3026
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015866084826174866,
+      "loss": 1.5626,
+      "step": 3027
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015863281165289027,
+      "loss": 1.4147,
+      "step": 3028
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015860476801894845,
+      "loss": 1.2598,
+      "step": 3029
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015857671736328323,
+      "loss": 1.4664,
+      "step": 3030
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015854865968925548,
+      "loss": 1.5001,
+      "step": 3031
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015852059500022695,
+      "loss": 1.1792,
+      "step": 3032
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015849252329956022,
+      "loss": 1.2919,
+      "step": 3033
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015846444459061864,
+      "loss": 1.1783,
+      "step": 3034
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015843635887676652,
+      "loss": 1.1875,
+      "step": 3035
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001584082661613689,
+      "loss": 1.6166,
+      "step": 3036
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015838016644779173,
+      "loss": 1.3771,
+      "step": 3037
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001583520597394018,
+      "loss": 1.124,
+      "step": 3038
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015832394603956663,
+      "loss": 1.4378,
+      "step": 3039
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015829582535165475,
+      "loss": 1.5858,
+      "step": 3040
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001582676976790354,
+      "loss": 1.4819,
+      "step": 3041
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015823956302507867,
+      "loss": 1.6958,
+      "step": 3042
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015821142139315554,
+      "loss": 1.1195,
+      "step": 3043
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001581832727866378,
+      "loss": 1.2436,
+      "step": 3044
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015815511720889804,
+      "loss": 1.4898,
+      "step": 3045
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015812695466330975,
+      "loss": 1.1942,
+      "step": 3046
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001580987851532472,
+      "loss": 1.32,
+      "step": 3047
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015807060868208554,
+      "loss": 1.5915,
+      "step": 3048
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001580424252532007,
+      "loss": 1.4278,
+      "step": 3049
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015801423486996952,
+      "loss": 1.3381,
+      "step": 3050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015798603753576957,
+      "loss": 1.2615,
+      "step": 3051
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015795783325397937,
+      "loss": 1.521,
+      "step": 3052
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015792962202797813,
+      "loss": 1.1599,
+      "step": 3053
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015790140386114608,
+      "loss": 1.267,
+      "step": 3054
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001578731787568641,
+      "loss": 1.3058,
+      "step": 3055
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000157844946718514,
+      "loss": 1.1848,
+      "step": 3056
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001578167077494784,
+      "loss": 1.4846,
+      "step": 3057
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015778846185314073,
+      "loss": 1.3616,
+      "step": 3058
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001577602090328853,
+      "loss": 1.4476,
+      "step": 3059
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015773194929209723,
+      "loss": 1.3119,
+      "step": 3060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015770368263416242,
+      "loss": 1.3033,
+      "step": 3061
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015767540906246763,
+      "loss": 1.4719,
+      "step": 3062
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015764712858040047,
+      "loss": 1.6018,
+      "step": 3063
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015761884119134937,
+      "loss": 1.6148,
+      "step": 3064
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001575905468987036,
+      "loss": 1.0995,
+      "step": 3065
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001575622457058532,
+      "loss": 1.078,
+      "step": 3066
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015753393761618907,
+      "loss": 1.3497,
+      "step": 3067
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015750562263310298,
+      "loss": 1.335,
+      "step": 3068
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015747730075998747,
+      "loss": 1.4207,
+      "step": 3069
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001574489720002359,
+      "loss": 1.5529,
+      "step": 3070
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001574206363572425,
+      "loss": 1.2633,
+      "step": 3071
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015739229383440227,
+      "loss": 1.6236,
+      "step": 3072
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001573639444351111,
+      "loss": 1.4017,
+      "step": 3073
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001573355881627657,
+      "loss": 1.5254,
+      "step": 3074
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001573072250207635,
+      "loss": 1.0216,
+      "step": 3075
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015727885501250286,
+      "loss": 1.2809,
+      "step": 3076
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015725047814138293,
+      "loss": 1.2915,
+      "step": 3077
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015722209441080369,
+      "loss": 1.3815,
+      "step": 3078
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015719370382416594,
+      "loss": 1.277,
+      "step": 3079
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015716530638487127,
+      "loss": 1.2204,
+      "step": 3080
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001571369020963221,
+      "loss": 1.4927,
+      "step": 3081
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015710849096192174,
+      "loss": 1.589,
+      "step": 3082
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001570800729850742,
+      "loss": 1.2743,
+      "step": 3083
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015705164816918446,
+      "loss": 1.5723,
+      "step": 3084
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015702321651765816,
+      "loss": 0.9142,
+      "step": 3085
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015699477803390188,
+      "loss": 1.1237,
+      "step": 3086
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015696633272132296,
+      "loss": 1.3807,
+      "step": 3087
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001569378805833296,
+      "loss": 1.6426,
+      "step": 3088
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001569094216233307,
+      "loss": 1.0396,
+      "step": 3089
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015688095584473615,
+      "loss": 1.5073,
+      "step": 3090
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015685248325095658,
+      "loss": 1.3379,
+      "step": 3091
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001568240038454034,
+      "loss": 1.4185,
+      "step": 3092
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015679551763148885,
+      "loss": 1.7129,
+      "step": 3093
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015676702461262602,
+      "loss": 1.4747,
+      "step": 3094
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015673852479222881,
+      "loss": 1.4618,
+      "step": 3095
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015671001817371192,
+      "loss": 1.367,
+      "step": 3096
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015668150476049085,
+      "loss": 1.6474,
+      "step": 3097
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015665298455598193,
+      "loss": 1.3613,
+      "step": 3098
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001566244575636023,
+      "loss": 1.6087,
+      "step": 3099
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015659592378676997,
+      "loss": 1.206,
+      "step": 3100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015656738322890367,
+      "loss": 1.3839,
+      "step": 3101
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015653883589342299,
+      "loss": 1.3835,
+      "step": 3102
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001565102817837483,
+      "loss": 1.2834,
+      "step": 3103
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015648172090330085,
+      "loss": 1.1711,
+      "step": 3104
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001564531532555026,
+      "loss": 1.447,
+      "step": 3105
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015642457884377646,
+      "loss": 1.2937,
+      "step": 3106
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015639599767154603,
+      "loss": 1.2276,
+      "step": 3107
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015636740974223575,
+      "loss": 1.348,
+      "step": 3108
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015633881505927087,
+      "loss": 1.1979,
+      "step": 3109
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015631021362607752,
+      "loss": 1.0952,
+      "step": 3110
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015628160544608251,
+      "loss": 1.1421,
+      "step": 3111
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015625299052271355,
+      "loss": 1.1293,
+      "step": 3112
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015622436885939913,
+      "loss": 1.645,
+      "step": 3113
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015619574045956856,
+      "loss": 1.759,
+      "step": 3114
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015616710532665194,
+      "loss": 1.4679,
+      "step": 3115
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015613846346408018,
+      "loss": 1.2362,
+      "step": 3116
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000156109814875285,
+      "loss": 0.8902,
+      "step": 3117
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015608115956369898,
+      "loss": 1.5422,
+      "step": 3118
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001560524975327554,
+      "loss": 1.2525,
+      "step": 3119
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015602382878588837,
+      "loss": 1.337,
+      "step": 3120
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001559951533265329,
+      "loss": 1.5652,
+      "step": 3121
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001559664711581247,
+      "loss": 1.469,
+      "step": 3122
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015593778228410033,
+      "loss": 1.3164,
+      "step": 3123
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015590908670789713,
+      "loss": 1.2847,
+      "step": 3124
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015588038443295328,
+      "loss": 1.5722,
+      "step": 3125
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015585167546270772,
+      "loss": 1.2805,
+      "step": 3126
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015582295980060023,
+      "loss": 1.0633,
+      "step": 3127
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015579423745007133,
+      "loss": 1.2545,
+      "step": 3128
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015576550841456247,
+      "loss": 1.2427,
+      "step": 3129
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015573677269751571,
+      "loss": 1.3396,
+      "step": 3130
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015570803030237414,
+      "loss": 1.4455,
+      "step": 3131
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001556792812325814,
+      "loss": 1.3528,
+      "step": 3132
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015565052549158215,
+      "loss": 1.4297,
+      "step": 3133
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001556217630828217,
+      "loss": 1.5171,
+      "step": 3134
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015559299400974623,
+      "loss": 1.4504,
+      "step": 3135
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015556421827580272,
+      "loss": 1.5423,
+      "step": 3136
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015553543588443894,
+      "loss": 1.2385,
+      "step": 3137
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015550664683910338,
+      "loss": 1.2624,
+      "step": 3138
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015547785114324549,
+      "loss": 1.3499,
+      "step": 3139
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015544904880031535,
+      "loss": 1.1432,
+      "step": 3140
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015542023981376395,
+      "loss": 1.3107,
+      "step": 3141
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015539142418704302,
+      "loss": 1.2867,
+      "step": 3142
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001553626019236051,
+      "loss": 0.9887,
+      "step": 3143
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015533377302690354,
+      "loss": 1.1137,
+      "step": 3144
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015530493750039244,
+      "loss": 1.0904,
+      "step": 3145
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015527609534752673,
+      "loss": 1.2372,
+      "step": 3146
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015524724657176218,
+      "loss": 1.4285,
+      "step": 3147
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015521839117655525,
+      "loss": 1.1312,
+      "step": 3148
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001551895291653633,
+      "loss": 1.5699,
+      "step": 3149
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015516066054164433,
+      "loss": 1.3067,
+      "step": 3150
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015513178530885733,
+      "loss": 1.348,
+      "step": 3151
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015510290347046192,
+      "loss": 1.5742,
+      "step": 3152
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015507401502991863,
+      "loss": 1.456,
+      "step": 3153
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015504511999068867,
+      "loss": 1.4316,
+      "step": 3154
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015501621835623412,
+      "loss": 1.5193,
+      "step": 3155
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015498731013001784,
+      "loss": 1.2989,
+      "step": 3156
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001549583953155034,
+      "loss": 1.1141,
+      "step": 3157
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015492947391615536,
+      "loss": 1.3138,
+      "step": 3158
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001549005459354388,
+      "loss": 1.5238,
+      "step": 3159
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015487161137681978,
+      "loss": 1.2748,
+      "step": 3160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001548426702437651,
+      "loss": 1.3414,
+      "step": 3161
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015481372253974232,
+      "loss": 1.5806,
+      "step": 3162
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015478476826821978,
+      "loss": 1.3554,
+      "step": 3163
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015475580743266667,
+      "loss": 1.0913,
+      "step": 3164
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001547268400365529,
+      "loss": 1.4317,
+      "step": 3165
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015469786608334927,
+      "loss": 1.3317,
+      "step": 3166
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001546688855765272,
+      "loss": 1.2285,
+      "step": 3167
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015463989851955903,
+      "loss": 1.2327,
+      "step": 3168
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015461090491591782,
+      "loss": 1.1849,
+      "step": 3169
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015458190476907746,
+      "loss": 1.3121,
+      "step": 3170
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001545528980825126,
+      "loss": 1.3097,
+      "step": 3171
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015452388485969864,
+      "loss": 1.2911,
+      "step": 3172
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015449486510411183,
+      "loss": 1.1898,
+      "step": 3173
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015446583881922914,
+      "loss": 1.6878,
+      "step": 3174
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015443680600852835,
+      "loss": 1.8455,
+      "step": 3175
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015440776667548806,
+      "loss": 1.3002,
+      "step": 3176
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015437872082358758,
+      "loss": 1.4431,
+      "step": 3177
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015434966845630704,
+      "loss": 1.2748,
+      "step": 3178
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001543206095771274,
+      "loss": 1.3792,
+      "step": 3179
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015429154418953025,
+      "loss": 1.4595,
+      "step": 3180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015426247229699812,
+      "loss": 1.3235,
+      "step": 3181
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015423339390301425,
+      "loss": 1.4175,
+      "step": 3182
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015420430901106266,
+      "loss": 1.4458,
+      "step": 3183
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015417521762462814,
+      "loss": 1.4339,
+      "step": 3184
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001541461197471963,
+      "loss": 1.5589,
+      "step": 3185
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015411701538225344,
+      "loss": 1.2764,
+      "step": 3186
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015408790453328675,
+      "loss": 1.3433,
+      "step": 3187
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015405878720378414,
+      "loss": 1.2762,
+      "step": 3188
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015402966339723425,
+      "loss": 1.4947,
+      "step": 3189
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015400053311712665,
+      "loss": 0.9838,
+      "step": 3190
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015397139636695145,
+      "loss": 1.5255,
+      "step": 3191
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015394225315019973,
+      "loss": 1.1883,
+      "step": 3192
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001539131034703633,
+      "loss": 1.3588,
+      "step": 3193
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015388394733093473,
+      "loss": 1.1985,
+      "step": 3194
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015385478473540728,
+      "loss": 1.5762,
+      "step": 3195
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015382561568727513,
+      "loss": 1.3104,
+      "step": 3196
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001537964401900332,
+      "loss": 1.3155,
+      "step": 3197
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015376725824717708,
+      "loss": 1.3486,
+      "step": 3198
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001537380698622032,
+      "loss": 1.3684,
+      "step": 3199
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015370887503860883,
+      "loss": 1.0818,
+      "step": 3200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015367967377989187,
+      "loss": 1.2858,
+      "step": 3201
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001536504660895511,
+      "loss": 1.6436,
+      "step": 3202
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015362125197108606,
+      "loss": 1.3863,
+      "step": 3203
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015359203142799695,
+      "loss": 1.6727,
+      "step": 3204
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015356280446378494,
+      "loss": 1.4078,
+      "step": 3205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015353357108195177,
+      "loss": 1.0938,
+      "step": 3206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001535043312860001,
+      "loss": 1.5165,
+      "step": 3207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015347508507943322,
+      "loss": 1.3235,
+      "step": 3208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001534458324657553,
+      "loss": 1.583,
+      "step": 3209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015341657344847127,
+      "loss": 1.1484,
+      "step": 3210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015338730803108675,
+      "loss": 1.4287,
+      "step": 3211
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015335803621710815,
+      "loss": 1.2791,
+      "step": 3212
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015332875801004275,
+      "loss": 1.316,
+      "step": 3213
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015329947341339843,
+      "loss": 1.3281,
+      "step": 3214
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015327018243068398,
+      "loss": 1.2948,
+      "step": 3215
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001532408850654089,
+      "loss": 1.5399,
+      "step": 3216
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001532115813210834,
+      "loss": 1.5733,
+      "step": 3217
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015318227120121852,
+      "loss": 1.5793,
+      "step": 3218
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015315295470932604,
+      "loss": 1.6116,
+      "step": 3219
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015312363184891858,
+      "loss": 1.3224,
+      "step": 3220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001530943026235094,
+      "loss": 1.5789,
+      "step": 3221
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015306496703661253,
+      "loss": 1.6248,
+      "step": 3222
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001530356250917429,
+      "loss": 1.1464,
+      "step": 3223
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015300627679241607,
+      "loss": 1.4158,
+      "step": 3224
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015297692214214843,
+      "loss": 1.5314,
+      "step": 3225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015294756114445706,
+      "loss": 1.4046,
+      "step": 3226
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015291819380285987,
+      "loss": 1.1156,
+      "step": 3227
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015288882012087548,
+      "loss": 1.734,
+      "step": 3228
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015285944010202335,
+      "loss": 1.243,
+      "step": 3229
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001528300537498236,
+      "loss": 1.5288,
+      "step": 3230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015280066106779716,
+      "loss": 1.455,
+      "step": 3231
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001527712620594657,
+      "loss": 1.5347,
+      "step": 3232
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015274185672835167,
+      "loss": 1.2557,
+      "step": 3233
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001527124450779783,
+      "loss": 1.3714,
+      "step": 3234
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015268302711186945,
+      "loss": 1.328,
+      "step": 3235
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015265360283354991,
+      "loss": 1.2806,
+      "step": 3236
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015262417224654517,
+      "loss": 1.3057,
+      "step": 3237
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015259473535438137,
+      "loss": 1.326,
+      "step": 3238
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015256529216058553,
+      "loss": 1.3689,
+      "step": 3239
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001525358426686854,
+      "loss": 1.3683,
+      "step": 3240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015250638688220944,
+      "loss": 1.3721,
+      "step": 3241
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001524769248046869,
+      "loss": 1.4914,
+      "step": 3242
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015244745643964777,
+      "loss": 1.3519,
+      "step": 3243
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001524179817906228,
+      "loss": 1.406,
+      "step": 3244
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015238850086114349,
+      "loss": 1.1671,
+      "step": 3245
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001523590136547421,
+      "loss": 1.4392,
+      "step": 3246
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015232952017495165,
+      "loss": 1.0627,
+      "step": 3247
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015230002042530586,
+      "loss": 1.2969,
+      "step": 3248
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001522705144093393,
+      "loss": 1.3127,
+      "step": 3249
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015224100213058716,
+      "loss": 1.5226,
+      "step": 3250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001522114835925855,
+      "loss": 1.6834,
+      "step": 3251
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015218195879887104,
+      "loss": 1.1388,
+      "step": 3252
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015215242775298127,
+      "loss": 1.4931,
+      "step": 3253
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001521228904584546,
+      "loss": 1.4521,
+      "step": 3254
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015209334691882983,
+      "loss": 1.0988,
+      "step": 3255
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015206379713764688,
+      "loss": 1.3532,
+      "step": 3256
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015203424111844615,
+      "loss": 1.4279,
+      "step": 3257
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015200467886476891,
+      "loss": 1.3949,
+      "step": 3258
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001519751103801572,
+      "loss": 1.4843,
+      "step": 3259
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015194553566815375,
+      "loss": 1.2809,
+      "step": 3260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015191595473230206,
+      "loss": 1.5412,
+      "step": 3261
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001518863675761463,
+      "loss": 1.8008,
+      "step": 3262
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015185677420323152,
+      "loss": 1.3628,
+      "step": 3263
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015182717461710349,
+      "loss": 1.2563,
+      "step": 3264
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001517975688213085,
+      "loss": 1.47,
+      "step": 3265
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015176795681939395,
+      "loss": 1.5426,
+      "step": 3266
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015173833861490768,
+      "loss": 1.3175,
+      "step": 3267
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001517087142113985,
+      "loss": 0.9873,
+      "step": 3268
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001516790836124158,
+      "loss": 1.4542,
+      "step": 3269
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015164944682150973,
+      "loss": 1.8196,
+      "step": 3270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015161980384223128,
+      "loss": 1.2403,
+      "step": 3271
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015159015467813201,
+      "loss": 1.573,
+      "step": 3272
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015156049933276447,
+      "loss": 1.5798,
+      "step": 3273
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001515308378096818,
+      "loss": 1.6203,
+      "step": 3274
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001515011701124378,
+      "loss": 0.9338,
+      "step": 3275
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015147149624458712,
+      "loss": 1.2052,
+      "step": 3276
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001514418162096852,
+      "loss": 1.5459,
+      "step": 3277
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001514121300112881,
+      "loss": 1.4504,
+      "step": 3278
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015138243765295268,
+      "loss": 1.4169,
+      "step": 3279
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015135273913823653,
+      "loss": 1.5374,
+      "step": 3280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015132303447069794,
+      "loss": 1.4127,
+      "step": 3281
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015129332365389605,
+      "loss": 1.2914,
+      "step": 3282
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015126360669139057,
+      "loss": 1.4114,
+      "step": 3283
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001512338835867421,
+      "loss": 1.2126,
+      "step": 3284
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015120415434351182,
+      "loss": 1.8072,
+      "step": 3285
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015117441896526183,
+      "loss": 1.3453,
+      "step": 3286
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015114467745555487,
+      "loss": 1.8423,
+      "step": 3287
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015111492981795434,
+      "loss": 1.4003,
+      "step": 3288
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015108517605602452,
+      "loss": 1.3795,
+      "step": 3289
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015105541617333032,
+      "loss": 1.7693,
+      "step": 3290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001510256501734374,
+      "loss": 1.1612,
+      "step": 3291
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015099587805991226,
+      "loss": 1.343,
+      "step": 3292
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001509660998363219,
+      "loss": 1.4869,
+      "step": 3293
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015093631550623432,
+      "loss": 1.1602,
+      "step": 3294
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015090652507321805,
+      "loss": 1.0441,
+      "step": 3295
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015087672854084246,
+      "loss": 1.486,
+      "step": 3296
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001508469259126776,
+      "loss": 1.8207,
+      "step": 3297
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015081711719229434,
+      "loss": 1.3765,
+      "step": 3298
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001507873023832641,
+      "loss": 1.2973,
+      "step": 3299
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015075748148915915,
+      "loss": 1.6596,
+      "step": 3300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015072765451355256,
+      "loss": 1.3173,
+      "step": 3301
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015069782146001802,
+      "loss": 1.4806,
+      "step": 3302
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015066798233212992,
+      "loss": 1.665,
+      "step": 3303
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001506381371334635,
+      "loss": 1.2501,
+      "step": 3304
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015060828586759456,
+      "loss": 1.2174,
+      "step": 3305
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015057842853809982,
+      "loss": 1.4624,
+      "step": 3306
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001505485651485566,
+      "loss": 1.3715,
+      "step": 3307
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015051869570254298,
+      "loss": 1.2282,
+      "step": 3308
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015048882020363778,
+      "loss": 1.4298,
+      "step": 3309
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001504589386554205,
+      "loss": 1.2536,
+      "step": 3310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001504290510614714,
+      "loss": 1.3614,
+      "step": 3311
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015039915742537144,
+      "loss": 1.4323,
+      "step": 3312
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001503692577507024,
+      "loss": 1.1121,
+      "step": 3313
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015033935204104666,
+      "loss": 1.3497,
+      "step": 3314
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015030944029998733,
+      "loss": 1.323,
+      "step": 3315
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015027952253110832,
+      "loss": 1.4661,
+      "step": 3316
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015024959873799425,
+      "loss": 1.4141,
+      "step": 3317
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015021966892423037,
+      "loss": 1.6461,
+      "step": 3318
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015018973309340275,
+      "loss": 1.3219,
+      "step": 3319
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001501597912490982,
+      "loss": 1.4718,
+      "step": 3320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001501298433949041,
+      "loss": 1.0252,
+      "step": 3321
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001500998895344087,
+      "loss": 1.2318,
+      "step": 3322
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015006992967120093,
+      "loss": 1.3266,
+      "step": 3323
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015003996380887045,
+      "loss": 1.5423,
+      "step": 3324
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015000999195100755,
+      "loss": 1.4169,
+      "step": 3325
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014998001410120328,
+      "loss": 1.367,
+      "step": 3326
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014995003026304954,
+      "loss": 1.3033,
+      "step": 3327
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014992004044013881,
+      "loss": 1.2836,
+      "step": 3328
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014989004463606423,
+      "loss": 1.4345,
+      "step": 3329
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014986004285441986,
+      "loss": 1.2745,
+      "step": 3330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001498300350988003,
+      "loss": 1.0814,
+      "step": 3331
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014980002137280092,
+      "loss": 1.2896,
+      "step": 3332
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014977000168001783,
+      "loss": 1.749,
+      "step": 3333
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014973997602404782,
+      "loss": 1.1628,
+      "step": 3334
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014970994440848844,
+      "loss": 1.4257,
+      "step": 3335
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001496799068369379,
+      "loss": 1.3261,
+      "step": 3336
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014964986331299516,
+      "loss": 1.4739,
+      "step": 3337
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014961981384025984,
+      "loss": 1.7945,
+      "step": 3338
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001495897584223324,
+      "loss": 1.6163,
+      "step": 3339
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014955969706281384,
+      "loss": 1.5942,
+      "step": 3340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014952962976530604,
+      "loss": 2.031,
+      "step": 3341
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014949955653341138,
+      "loss": 1.0447,
+      "step": 3342
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014946947737073322,
+      "loss": 1.3473,
+      "step": 3343
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014943939228087542,
+      "loss": 1.336,
+      "step": 3344
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014940930126744264,
+      "loss": 1.4257,
+      "step": 3345
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014937920433404021,
+      "loss": 1.4069,
+      "step": 3346
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014934910148427425,
+      "loss": 1.4728,
+      "step": 3347
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014931899272175146,
+      "loss": 1.4171,
+      "step": 3348
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014928887805007937,
+      "loss": 1.5982,
+      "step": 3349
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001492587574728661,
+      "loss": 1.3607,
+      "step": 3350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014922863099372065,
+      "loss": 1.1828,
+      "step": 3351
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001491984986162525,
+      "loss": 1.2668,
+      "step": 3352
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014916836034407202,
+      "loss": 1.5283,
+      "step": 3353
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014913821618079025,
+      "loss": 1.0926,
+      "step": 3354
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014910806613001884,
+      "loss": 1.2817,
+      "step": 3355
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014907791019537028,
+      "loss": 1.3711,
+      "step": 3356
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014904774838045762,
+      "loss": 1.3322,
+      "step": 3357
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001490175806888948,
+      "loss": 0.9086,
+      "step": 3358
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014898740712429627,
+      "loss": 1.4353,
+      "step": 3359
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014895722769027732,
+      "loss": 1.1907,
+      "step": 3360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001489270423904539,
+      "loss": 1.5591,
+      "step": 3361
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001488968512284426,
+      "loss": 1.4458,
+      "step": 3362
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014886665420786084,
+      "loss": 1.5406,
+      "step": 3363
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014883645133232664,
+      "loss": 1.8418,
+      "step": 3364
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014880624260545875,
+      "loss": 1.5401,
+      "step": 3365
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014877602803087662,
+      "loss": 1.6643,
+      "step": 3366
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001487458076122004,
+      "loss": 1.769,
+      "step": 3367
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014871558135305101,
+      "loss": 1.94,
+      "step": 3368
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014868534925704992,
+      "loss": 1.3099,
+      "step": 3369
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014865511132781945,
+      "loss": 1.1294,
+      "step": 3370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014862486756898253,
+      "loss": 1.245,
+      "step": 3371
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014859461798416278,
+      "loss": 1.8105,
+      "step": 3372
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014856436257698459,
+      "loss": 1.2786,
+      "step": 3373
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014853410135107299,
+      "loss": 1.0868,
+      "step": 3374
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014850383431005377,
+      "loss": 1.5344,
+      "step": 3375
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014847356145755331,
+      "loss": 1.5565,
+      "step": 3376
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014844328279719877,
+      "loss": 1.3477,
+      "step": 3377
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014841299833261802,
+      "loss": 1.6843,
+      "step": 3378
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014838270806743952,
+      "loss": 1.2647,
+      "step": 3379
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014835241200529252,
+      "loss": 0.7982,
+      "step": 3380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000148322110149807,
+      "loss": 1.5868,
+      "step": 3381
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014829180250461348,
+      "loss": 1.3364,
+      "step": 3382
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014826148907334333,
+      "loss": 1.4151,
+      "step": 3383
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014823116985962856,
+      "loss": 1.7731,
+      "step": 3384
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014820084486710179,
+      "loss": 1.6994,
+      "step": 3385
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014817051409939645,
+      "loss": 1.1901,
+      "step": 3386
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014814017756014663,
+      "loss": 1.4545,
+      "step": 3387
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014810983525298708,
+      "loss": 1.247,
+      "step": 3388
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014807948718155328,
+      "loss": 1.0816,
+      "step": 3389
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001480491333494814,
+      "loss": 1.5892,
+      "step": 3390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001480187737604082,
+      "loss": 1.0929,
+      "step": 3391
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014798840841797126,
+      "loss": 1.2527,
+      "step": 3392
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001479580373258088,
+      "loss": 1.3145,
+      "step": 3393
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014792766048755975,
+      "loss": 1.2737,
+      "step": 3394
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014789727790686364,
+      "loss": 1.3485,
+      "step": 3395
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014786688958736086,
+      "loss": 1.0681,
+      "step": 3396
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014783649553269228,
+      "loss": 1.3538,
+      "step": 3397
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014780609574649962,
+      "loss": 1.375,
+      "step": 3398
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001477756902324252,
+      "loss": 1.3474,
+      "step": 3399
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014774527899411206,
+      "loss": 1.304,
+      "step": 3400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014771486203520392,
+      "loss": 1.2653,
+      "step": 3401
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001476844393593452,
+      "loss": 1.4167,
+      "step": 3402
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014765401097018097,
+      "loss": 1.6493,
+      "step": 3403
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000147623576871357,
+      "loss": 1.0299,
+      "step": 3404
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001475931370665198,
+      "loss": 1.3599,
+      "step": 3405
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001475626915593164,
+      "loss": 1.3616,
+      "step": 3406
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014753224035339474,
+      "loss": 1.2853,
+      "step": 3407
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001475017834524033,
+      "loss": 1.1124,
+      "step": 3408
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014747132085999118,
+      "loss": 1.0193,
+      "step": 3409
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014744085257980839,
+      "loss": 1.6685,
+      "step": 3410
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014741037861550538,
+      "loss": 1.1634,
+      "step": 3411
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014737989897073344,
+      "loss": 1.1597,
+      "step": 3412
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014734941364914444,
+      "loss": 1.1783,
+      "step": 3413
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000147318922654391,
+      "loss": 1.3211,
+      "step": 3414
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014728842599012645,
+      "loss": 1.0435,
+      "step": 3415
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001472579236600046,
+      "loss": 1.4357,
+      "step": 3416
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014722741566768023,
+      "loss": 1.1773,
+      "step": 3417
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014719690201680862,
+      "loss": 1.6188,
+      "step": 3418
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001471663827110457,
+      "loss": 1.1617,
+      "step": 3419
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014713585775404816,
+      "loss": 1.2125,
+      "step": 3420
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014710532714947341,
+      "loss": 1.2351,
+      "step": 3421
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001470747909009794,
+      "loss": 1.884,
+      "step": 3422
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001470442490122248,
+      "loss": 1.2432,
+      "step": 3423
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014701370148686906,
+      "loss": 1.1854,
+      "step": 3424
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001469831483285722,
+      "loss": 1.5765,
+      "step": 3425
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001469525895409949,
+      "loss": 1.0732,
+      "step": 3426
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014692202512779862,
+      "loss": 1.4795,
+      "step": 3427
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014689145509264544,
+      "loss": 1.5864,
+      "step": 3428
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014686087943919805,
+      "loss": 1.1953,
+      "step": 3429
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014683029817111985,
+      "loss": 1.1202,
+      "step": 3430
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014679971129207502,
+      "loss": 1.1607,
+      "step": 3431
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014676911880572827,
+      "loss": 1.1994,
+      "step": 3432
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000146738520715745,
+      "loss": 1.5665,
+      "step": 3433
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001467079170257914,
+      "loss": 1.3552,
+      "step": 3434
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001466773077395342,
+      "loss": 1.2084,
+      "step": 3435
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014664669286064083,
+      "loss": 1.2429,
+      "step": 3436
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014661607239277943,
+      "loss": 1.3253,
+      "step": 3437
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014658544633961882,
+      "loss": 1.369,
+      "step": 3438
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001465548147048284,
+      "loss": 1.4008,
+      "step": 3439
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001465241774920783,
+      "loss": 1.5082,
+      "step": 3440
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001464935347050394,
+      "loss": 1.2523,
+      "step": 3441
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014646288634738303,
+      "loss": 1.1272,
+      "step": 3442
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001464322324227814,
+      "loss": 1.1133,
+      "step": 3443
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001464015729349073,
+      "loss": 1.259,
+      "step": 3444
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014637090788743423,
+      "loss": 1.1742,
+      "step": 3445
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001463402372840362,
+      "loss": 1.1604,
+      "step": 3446
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014630956112838812,
+      "loss": 1.5584,
+      "step": 3447
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001462788794241654,
+      "loss": 1.2101,
+      "step": 3448
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014624819217504416,
+      "loss": 1.2444,
+      "step": 3449
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014621749938470123,
+      "loss": 1.2505,
+      "step": 3450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000146186801056814,
+      "loss": 1.1321,
+      "step": 3451
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014615609719506066,
+      "loss": 1.2277,
+      "step": 3452
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014612538780311995,
+      "loss": 1.2111,
+      "step": 3453
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001460946728846713,
+      "loss": 1.65,
+      "step": 3454
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014606395244339484,
+      "loss": 1.1408,
+      "step": 3455
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001460332264829713,
+      "loss": 1.3209,
+      "step": 3456
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014600249500708217,
+      "loss": 1.1763,
+      "step": 3457
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014597175801940948,
+      "loss": 1.3315,
+      "step": 3458
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000145941015523636,
+      "loss": 1.2567,
+      "step": 3459
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014591026752344512,
+      "loss": 1.3126,
+      "step": 3460
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014587951402252096,
+      "loss": 1.3388,
+      "step": 3461
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001458487550245482,
+      "loss": 1.39,
+      "step": 3462
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014581799053321225,
+      "loss": 1.2151,
+      "step": 3463
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014578722055219912,
+      "loss": 1.0948,
+      "step": 3464
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014575644508519554,
+      "loss": 1.586,
+      "step": 3465
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014572566413588884,
+      "loss": 1.2468,
+      "step": 3466
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014569487770796707,
+      "loss": 0.7208,
+      "step": 3467
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014566408580511894,
+      "loss": 1.7631,
+      "step": 3468
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001456332884310337,
+      "loss": 1.1592,
+      "step": 3469
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014560248558940134,
+      "loss": 1.0552,
+      "step": 3470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014557167728391254,
+      "loss": 1.6353,
+      "step": 3471
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014554086351825857,
+      "loss": 1.3956,
+      "step": 3472
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001455100442961314,
+      "loss": 1.579,
+      "step": 3473
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014547921962122359,
+      "loss": 1.2718,
+      "step": 3474
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014544838949722842,
+      "loss": 1.3857,
+      "step": 3475
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014541755392783983,
+      "loss": 1.3337,
+      "step": 3476
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001453867129167523,
+      "loss": 1.6477,
+      "step": 3477
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014535586646766114,
+      "loss": 1.4334,
+      "step": 3478
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001453250145842621,
+      "loss": 1.5085,
+      "step": 3479
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001452941572702518,
+      "loss": 1.064,
+      "step": 3480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014526329452932735,
+      "loss": 1.4575,
+      "step": 3481
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014523242636518661,
+      "loss": 1.3246,
+      "step": 3482
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014520155278152798,
+      "loss": 1.316,
+      "step": 3483
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014517067378205063,
+      "loss": 1.4815,
+      "step": 3484
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001451397893704543,
+      "loss": 1.2793,
+      "step": 3485
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001451088995504394,
+      "loss": 1.2229,
+      "step": 3486
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014507800432570698,
+      "loss": 1.2604,
+      "step": 3487
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014504710369995876,
+      "loss": 1.2205,
+      "step": 3488
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001450161976768971,
+      "loss": 1.0083,
+      "step": 3489
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000144985286260225,
+      "loss": 1.2558,
+      "step": 3490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014495436945364615,
+      "loss": 1.3796,
+      "step": 3491
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014492344726086475,
+      "loss": 1.3479,
+      "step": 3492
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001448925196855858,
+      "loss": 0.9915,
+      "step": 3493
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001448615867315149,
+      "loss": 1.1641,
+      "step": 3494
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014483064840235826,
+      "loss": 1.4736,
+      "step": 3495
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001447997047018227,
+      "loss": 1.1814,
+      "step": 3496
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014476875563361577,
+      "loss": 1.4194,
+      "step": 3497
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014473780120144572,
+      "loss": 1.3017,
+      "step": 3498
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014470684140902119,
+      "loss": 1.1865,
+      "step": 3499
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014467587626005172,
+      "loss": 1.7005,
+      "step": 3500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001446449057582474,
+      "loss": 1.1847,
+      "step": 3501
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014461392990731893,
+      "loss": 1.5513,
+      "step": 3502
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014458294871097768,
+      "loss": 1.3512,
+      "step": 3503
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014455196217293565,
+      "loss": 1.2511,
+      "step": 3504
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014452097029690554,
+      "loss": 1.5593,
+      "step": 3505
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014448997308660056,
+      "loss": 1.3966,
+      "step": 3506
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001444589705457347,
+      "loss": 1.3268,
+      "step": 3507
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001444279626780225,
+      "loss": 1.4613,
+      "step": 3508
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014439694948717918,
+      "loss": 1.1566,
+      "step": 3509
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014436593097692053,
+      "loss": 1.4854,
+      "step": 3510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001443349071509631,
+      "loss": 1.2112,
+      "step": 3511
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014430387801302397,
+      "loss": 1.2125,
+      "step": 3512
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001442728435668209,
+      "loss": 1.5443,
+      "step": 3513
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014424180381607225,
+      "loss": 1.3382,
+      "step": 3514
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001442107587644971,
+      "loss": 1.6266,
+      "step": 3515
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014417970841581505,
+      "loss": 1.3904,
+      "step": 3516
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014414865277374644,
+      "loss": 1.1035,
+      "step": 3517
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014411759184201214,
+      "loss": 1.1378,
+      "step": 3518
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001440865256243338,
+      "loss": 1.6925,
+      "step": 3519
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014405545412443356,
+      "loss": 1.2904,
+      "step": 3520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014402437734603422,
+      "loss": 1.3978,
+      "step": 3521
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001439932952928593,
+      "loss": 1.2264,
+      "step": 3522
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014396220796863288,
+      "loss": 1.7149,
+      "step": 3523
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014393111537707966,
+      "loss": 1.598,
+      "step": 3524
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014390001752192502,
+      "loss": 1.4066,
+      "step": 3525
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014386891440689491,
+      "loss": 1.3764,
+      "step": 3526
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000143837806035716,
+      "loss": 1.3848,
+      "step": 3527
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014380669241211548,
+      "loss": 1.6557,
+      "step": 3528
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014377557353982128,
+      "loss": 1.6056,
+      "step": 3529
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001437444494225618,
+      "loss": 1.1393,
+      "step": 3530
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014371332006406635,
+      "loss": 1.389,
+      "step": 3531
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014368218546806453,
+      "loss": 1.232,
+      "step": 3532
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001436510456382868,
+      "loss": 1.4764,
+      "step": 3533
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014361990057846415,
+      "loss": 1.3112,
+      "step": 3534
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014358875029232823,
+      "loss": 0.9274,
+      "step": 3535
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014355759478361133,
+      "loss": 1.617,
+      "step": 3536
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001435264340560463,
+      "loss": 1.042,
+      "step": 3537
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001434952681133667,
+      "loss": 1.1648,
+      "step": 3538
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014346409695930665,
+      "loss": 1.0106,
+      "step": 3539
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014343292059760092,
+      "loss": 1.2698,
+      "step": 3540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001434017390319849,
+      "loss": 1.3431,
+      "step": 3541
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014337055226619464,
+      "loss": 1.2062,
+      "step": 3542
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014333936030396673,
+      "loss": 1.1222,
+      "step": 3543
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014330816314903845,
+      "loss": 1.4134,
+      "step": 3544
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014327696080514772,
+      "loss": 1.1105,
+      "step": 3545
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014324575327603298,
+      "loss": 1.3102,
+      "step": 3546
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001432145405654334,
+      "loss": 1.4007,
+      "step": 3547
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014318332267708868,
+      "loss": 1.5443,
+      "step": 3548
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014315209961473924,
+      "loss": 1.3064,
+      "step": 3549
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014312087138212607,
+      "loss": 1.6409,
+      "step": 3550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014308963798299074,
+      "loss": 1.0123,
+      "step": 3551
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001430583994210755,
+      "loss": 1.1135,
+      "step": 3552
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001430271557001232,
+      "loss": 1.4356,
+      "step": 3553
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014299590682387726,
+      "loss": 1.3535,
+      "step": 3554
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014296465279608182,
+      "loss": 1.6892,
+      "step": 3555
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014293339362048151,
+      "loss": 1.2311,
+      "step": 3556
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001429021293008217,
+      "loss": 1.5283,
+      "step": 3557
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014287085984084834,
+      "loss": 1.3679,
+      "step": 3558
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014283958524430792,
+      "loss": 1.2173,
+      "step": 3559
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014280830551494756,
+      "loss": 1.3936,
+      "step": 3560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001427770206565152,
+      "loss": 1.5758,
+      "step": 3561
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014274573067275906,
+      "loss": 1.3371,
+      "step": 3562
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014271443556742824,
+      "loss": 1.0995,
+      "step": 3563
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014268313534427233,
+      "loss": 1.0335,
+      "step": 3564
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001426518300070416,
+      "loss": 1.2742,
+      "step": 3565
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014262051955948684,
+      "loss": 1.4932,
+      "step": 3566
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014258920400535952,
+      "loss": 1.3978,
+      "step": 3567
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014255788334841175,
+      "loss": 0.8532,
+      "step": 3568
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014252655759239613,
+      "loss": 1.5336,
+      "step": 3569
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014249522674106605,
+      "loss": 1.3502,
+      "step": 3570
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014246389079817536,
+      "loss": 1.1474,
+      "step": 3571
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001424325497674786,
+      "loss": 1.3446,
+      "step": 3572
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001424012036527308,
+      "loss": 1.3562,
+      "step": 3573
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014236985245768782,
+      "loss": 1.3621,
+      "step": 3574
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014233849618610592,
+      "loss": 1.4407,
+      "step": 3575
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001423071348417421,
+      "loss": 1.4116,
+      "step": 3576
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014227576842835383,
+      "loss": 1.1914,
+      "step": 3577
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014224439694969938,
+      "loss": 1.3773,
+      "step": 3578
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014221302040953745,
+      "loss": 1.404,
+      "step": 3579
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014218163881162742,
+      "loss": 1.3547,
+      "step": 3580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014215025215972928,
+      "loss": 1.404,
+      "step": 3581
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014211886045760368,
+      "loss": 1.2706,
+      "step": 3582
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014208746370901172,
+      "loss": 1.3334,
+      "step": 3583
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001420560619177152,
+      "loss": 1.5396,
+      "step": 3584
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014202465508747662,
+      "loss": 1.4436,
+      "step": 3585
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001419932432220589,
+      "loss": 1.5209,
+      "step": 3586
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014196182632522566,
+      "loss": 1.1587,
+      "step": 3587
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014193040440074113,
+      "loss": 1.3372,
+      "step": 3588
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014189897745237013,
+      "loss": 1.3459,
+      "step": 3589
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014186754548387807,
+      "loss": 1.347,
+      "step": 3590
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014183610849903092,
+      "loss": 1.0089,
+      "step": 3591
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014180466650159538,
+      "loss": 1.4549,
+      "step": 3592
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014177321949533864,
+      "loss": 1.3023,
+      "step": 3593
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014174176748402848,
+      "loss": 1.2168,
+      "step": 3594
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014171031047143336,
+      "loss": 1.5026,
+      "step": 3595
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014167884846132226,
+      "loss": 1.2332,
+      "step": 3596
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014164738145746485,
+      "loss": 1.2204,
+      "step": 3597
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014161590946363132,
+      "loss": 1.5872,
+      "step": 3598
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014158443248359247,
+      "loss": 1.2045,
+      "step": 3599
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014155295052111974,
+      "loss": 1.596,
+      "step": 3600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014152146357998514,
+      "loss": 1.2506,
+      "step": 3601
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014148997166396122,
+      "loss": 1.2225,
+      "step": 3602
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014145847477682122,
+      "loss": 1.2684,
+      "step": 3603
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014142697292233894,
+      "loss": 1.3057,
+      "step": 3604
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014139546610428876,
+      "loss": 1.4812,
+      "step": 3605
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014136395432644565,
+      "loss": 1.1432,
+      "step": 3606
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014133243759258522,
+      "loss": 1.467,
+      "step": 3607
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014130091590648362,
+      "loss": 1.1282,
+      "step": 3608
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014126938927191767,
+      "loss": 1.381,
+      "step": 3609
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014123785769266464,
+      "loss": 1.07,
+      "step": 3610
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014120632117250258,
+      "loss": 1.2378,
+      "step": 3611
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014117477971520995,
+      "loss": 1.61,
+      "step": 3612
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014114323332456593,
+      "loss": 1.482,
+      "step": 3613
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014111168200435024,
+      "loss": 1.2907,
+      "step": 3614
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001410801257583432,
+      "loss": 1.2198,
+      "step": 3615
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001410485645903257,
+      "loss": 1.3635,
+      "step": 3616
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014101699850407924,
+      "loss": 1.2425,
+      "step": 3617
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014098542750338596,
+      "loss": 1.3695,
+      "step": 3618
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014095385159202848,
+      "loss": 1.1703,
+      "step": 3619
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014092227077379006,
+      "loss": 1.4628,
+      "step": 3620
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014089068505245458,
+      "loss": 1.3937,
+      "step": 3621
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014085909443180646,
+      "loss": 1.3862,
+      "step": 3622
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014082749891563076,
+      "loss": 1.0971,
+      "step": 3623
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014079589850771304,
+      "loss": 1.5372,
+      "step": 3624
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014076429321183956,
+      "loss": 1.467,
+      "step": 3625
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014073268303179708,
+      "loss": 1.6554,
+      "step": 3626
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014070106797137294,
+      "loss": 1.2382,
+      "step": 3627
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014066944803435512,
+      "loss": 1.3502,
+      "step": 3628
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014063782322453215,
+      "loss": 1.1938,
+      "step": 3629
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001406061935456932,
+      "loss": 0.9622,
+      "step": 3630
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001405745590016279,
+      "loss": 1.4464,
+      "step": 3631
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014054291959612655,
+      "loss": 1.3591,
+      "step": 3632
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014051127533298009,
+      "loss": 1.4329,
+      "step": 3633
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001404796262159799,
+      "loss": 1.2738,
+      "step": 3634
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014044797224891807,
+      "loss": 1.342,
+      "step": 3635
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001404163134355872,
+      "loss": 1.1925,
+      "step": 3636
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014038464977978045,
+      "loss": 1.5994,
+      "step": 3637
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014035298128529166,
+      "loss": 1.2645,
+      "step": 3638
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014032130795591516,
+      "loss": 1.684,
+      "step": 3639
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014028962979544584,
+      "loss": 1.584,
+      "step": 3640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014025794680767926,
+      "loss": 1.2425,
+      "step": 3641
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014022625899641155,
+      "loss": 1.3286,
+      "step": 3642
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014019456636543934,
+      "loss": 1.2656,
+      "step": 3643
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014016286891855984,
+      "loss": 1.2717,
+      "step": 3644
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014013116665957093,
+      "loss": 1.2147,
+      "step": 3645
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000140099459592271,
+      "loss": 1.2506,
+      "step": 3646
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014006774772045902,
+      "loss": 1.4994,
+      "step": 3647
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014003603104793455,
+      "loss": 1.0029,
+      "step": 3648
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014000430957849773,
+      "loss": 1.2681,
+      "step": 3649
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013997258331594923,
+      "loss": 1.3432,
+      "step": 3650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013994085226409037,
+      "loss": 1.5451,
+      "step": 3651
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000139909116426723,
+      "loss": 1.6486,
+      "step": 3652
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001398773758076495,
+      "loss": 0.9838,
+      "step": 3653
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001398456304106729,
+      "loss": 1.3134,
+      "step": 3654
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013981388023959676,
+      "loss": 1.6579,
+      "step": 3655
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013978212529822527,
+      "loss": 1.2129,
+      "step": 3656
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013975036559036309,
+      "loss": 1.2027,
+      "step": 3657
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013971860111981548,
+      "loss": 1.4756,
+      "step": 3658
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013968683189038842,
+      "loss": 1.4427,
+      "step": 3659
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013965505790588823,
+      "loss": 1.4653,
+      "step": 3660
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013962327917012188,
+      "loss": 1.2103,
+      "step": 3661
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013959149568689705,
+      "loss": 1.3073,
+      "step": 3662
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013955970746002178,
+      "loss": 1.2438,
+      "step": 3663
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001395279144933048,
+      "loss": 1.3366,
+      "step": 3664
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013949611679055538,
+      "loss": 1.4389,
+      "step": 3665
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001394643143555834,
+      "loss": 1.333,
+      "step": 3666
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001394325071921992,
+      "loss": 1.7077,
+      "step": 3667
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013940069530421375,
+      "loss": 1.3661,
+      "step": 3668
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013936887869543866,
+      "loss": 1.6316,
+      "step": 3669
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000139337057369686,
+      "loss": 1.1866,
+      "step": 3670
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001393052313307684,
+      "loss": 1.3444,
+      "step": 3671
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013927340058249914,
+      "loss": 1.2058,
+      "step": 3672
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000139241565128692,
+      "loss": 1.5855,
+      "step": 3673
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013920972497316128,
+      "loss": 1.3539,
+      "step": 3674
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013917788011972207,
+      "loss": 1.5936,
+      "step": 3675
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013914603057218968,
+      "loss": 1.2508,
+      "step": 3676
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013911417633438025,
+      "loss": 1.0345,
+      "step": 3677
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013908231741011036,
+      "loss": 1.3405,
+      "step": 3678
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013905045380319721,
+      "loss": 1.2416,
+      "step": 3679
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013901858551745852,
+      "loss": 1.1314,
+      "step": 3680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001389867125567126,
+      "loss": 1.5837,
+      "step": 3681
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013895483492477827,
+      "loss": 1.562,
+      "step": 3682
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013892295262547496,
+      "loss": 1.2776,
+      "step": 3683
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013889106566262267,
+      "loss": 1.3385,
+      "step": 3684
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013885917404004192,
+      "loss": 1.2085,
+      "step": 3685
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013882727776155378,
+      "loss": 1.4026,
+      "step": 3686
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001387953768309799,
+      "loss": 1.2376,
+      "step": 3687
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001387634712521425,
+      "loss": 1.4921,
+      "step": 3688
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013873156102886438,
+      "loss": 1.2929,
+      "step": 3689
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013869964616496878,
+      "loss": 1.5756,
+      "step": 3690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013866772666427962,
+      "loss": 0.9848,
+      "step": 3691
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013863580253062137,
+      "loss": 1.3773,
+      "step": 3692
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013860387376781895,
+      "loss": 1.3882,
+      "step": 3693
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001385719403796979,
+      "loss": 1.5332,
+      "step": 3694
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013854000237008437,
+      "loss": 1.137,
+      "step": 3695
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013850805974280496,
+      "loss": 1.734,
+      "step": 3696
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001384761125016869,
+      "loss": 1.3658,
+      "step": 3697
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013844416065055795,
+      "loss": 1.7039,
+      "step": 3698
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001384122041932464,
+      "loss": 1.613,
+      "step": 3699
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001383802431335811,
+      "loss": 1.2716,
+      "step": 3700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013834827747539148,
+      "loss": 1.4163,
+      "step": 3701
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001383163072225075,
+      "loss": 0.9463,
+      "step": 3702
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013828433237875968,
+      "loss": 1.1282,
+      "step": 3703
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013825235294797903,
+      "loss": 1.1243,
+      "step": 3704
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013822036893399725,
+      "loss": 1.2957,
+      "step": 3705
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001381883803406464,
+      "loss": 1.6234,
+      "step": 3706
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013815638717175926,
+      "loss": 1.451,
+      "step": 3707
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001381243894311691,
+      "loss": 1.4668,
+      "step": 3708
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013809238712270968,
+      "loss": 1.3337,
+      "step": 3709
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013806038025021534,
+      "loss": 1.1799,
+      "step": 3710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013802836881752107,
+      "loss": 1.4155,
+      "step": 3711
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001379963528284622,
+      "loss": 1.4282,
+      "step": 3712
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001379643322868748,
+      "loss": 1.3677,
+      "step": 3713
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013793230719659534,
+      "loss": 1.2832,
+      "step": 3714
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000137900277561461,
+      "loss": 1.2002,
+      "step": 3715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013786824338530936,
+      "loss": 1.2149,
+      "step": 3716
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013783620467197854,
+      "loss": 1.2595,
+      "step": 3717
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001378041614253073,
+      "loss": 1.2739,
+      "step": 3718
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013777211364913492,
+      "loss": 1.1181,
+      "step": 3719
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013774006134730117,
+      "loss": 1.8458,
+      "step": 3720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013770800452364638,
+      "loss": 1.3057,
+      "step": 3721
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001376759431820115,
+      "loss": 1.2094,
+      "step": 3722
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013764387732623787,
+      "loss": 1.3076,
+      "step": 3723
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013761180696016752,
+      "loss": 1.1516,
+      "step": 3724
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001375797320876429,
+      "loss": 1.1913,
+      "step": 3725
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013754765271250713,
+      "loss": 1.7149,
+      "step": 3726
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013751556883860374,
+      "loss": 1.4114,
+      "step": 3727
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013748348046977685,
+      "loss": 1.1101,
+      "step": 3728
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001374513876098712,
+      "loss": 1.5315,
+      "step": 3729
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001374192902627319,
+      "loss": 1.1116,
+      "step": 3730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013738718843220475,
+      "loss": 1.2817,
+      "step": 3731
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013735508212213597,
+      "loss": 1.3279,
+      "step": 3732
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013732297133637247,
+      "loss": 1.2765,
+      "step": 3733
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001372908560787615,
+      "loss": 1.4961,
+      "step": 3734
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013725873635315097,
+      "loss": 1.3649,
+      "step": 3735
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013722661216338937,
+      "loss": 1.4422,
+      "step": 3736
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013719448351332556,
+      "loss": 0.9595,
+      "step": 3737
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001371623504068091,
+      "loss": 1.1714,
+      "step": 3738
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013713021284768998,
+      "loss": 1.0266,
+      "step": 3739
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001370980708398187,
+      "loss": 1.3709,
+      "step": 3740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001370659243870465,
+      "loss": 1.7273,
+      "step": 3741
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001370337734932249,
+      "loss": 1.4938,
+      "step": 3742
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013700161816220606,
+      "loss": 1.6741,
+      "step": 3743
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001369694583978427,
+      "loss": 1.4668,
+      "step": 3744
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013693729420398803,
+      "loss": 1.4278,
+      "step": 3745
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001369051255844958,
+      "loss": 1.4281,
+      "step": 3746
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013687295254322023,
+      "loss": 1.263,
+      "step": 3747
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001368407750840162,
+      "loss": 1.1916,
+      "step": 3748
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013680859321073906,
+      "loss": 1.5783,
+      "step": 3749
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013677640692724463,
+      "loss": 1.2653,
+      "step": 3750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001367442162373893,
+      "loss": 1.1858,
+      "step": 3751
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013671202114503002,
+      "loss": 1.3124,
+      "step": 3752
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013667982165402423,
+      "loss": 1.2522,
+      "step": 3753
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013664761776822993,
+      "loss": 1.1393,
+      "step": 3754
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013661540949150562,
+      "loss": 1.4973,
+      "step": 3755
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013658319682771032,
+      "loss": 1.3603,
+      "step": 3756
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013655097978070355,
+      "loss": 0.9658,
+      "step": 3757
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013651875835434544,
+      "loss": 1.4345,
+      "step": 3758
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001364865325524966,
+      "loss": 0.9678,
+      "step": 3759
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013645430237901814,
+      "loss": 1.1777,
+      "step": 3760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013642206783777173,
+      "loss": 1.2886,
+      "step": 3761
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001363898289326195,
+      "loss": 1.6011,
+      "step": 3762
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013635758566742423,
+      "loss": 1.6081,
+      "step": 3763
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013632533804604907,
+      "loss": 1.0299,
+      "step": 3764
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013629308607235782,
+      "loss": 0.9626,
+      "step": 3765
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013626082975021472,
+      "loss": 1.1111,
+      "step": 3766
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013622856908348453,
+      "loss": 1.3011,
+      "step": 3767
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001361963040760326,
+      "loss": 1.2626,
+      "step": 3768
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001361640347317248,
+      "loss": 1.2461,
+      "step": 3769
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013613176105442735,
+      "loss": 1.3766,
+      "step": 3770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001360994830480072,
+      "loss": 1.5069,
+      "step": 3771
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013606720071633176,
+      "loss": 1.156,
+      "step": 3772
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001360349140632689,
+      "loss": 1.244,
+      "step": 3773
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013600262309268701,
+      "loss": 1.7327,
+      "step": 3774
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013597032780845506,
+      "loss": 1.1363,
+      "step": 3775
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013593802821444253,
+      "loss": 0.8319,
+      "step": 3776
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013590572431451935,
+      "loss": 1.3596,
+      "step": 3777
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000135873416112556,
+      "loss": 1.2215,
+      "step": 3778
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001358411036124236,
+      "loss": 0.9191,
+      "step": 3779
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001358087868179935,
+      "loss": 0.8912,
+      "step": 3780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013577646573313782,
+      "loss": 1.075,
+      "step": 3781
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013574414036172914,
+      "loss": 1.2472,
+      "step": 3782
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013571181070764043,
+      "loss": 1.3093,
+      "step": 3783
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013567947677474532,
+      "loss": 1.5676,
+      "step": 3784
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001356471385669179,
+      "loss": 1.222,
+      "step": 3785
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013561479608803275,
+      "loss": 1.4303,
+      "step": 3786
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000135582449341965,
+      "loss": 1.2541,
+      "step": 3787
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013555009833259024,
+      "loss": 1.2702,
+      "step": 3788
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013551774306378467,
+      "loss": 1.2528,
+      "step": 3789
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013548538353942482,
+      "loss": 1.3338,
+      "step": 3790
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013545301976338793,
+      "loss": 1.6949,
+      "step": 3791
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013542065173955162,
+      "loss": 1.5019,
+      "step": 3792
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013538827947179413,
+      "loss": 1.406,
+      "step": 3793
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013535590296399406,
+      "loss": 1.2634,
+      "step": 3794
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013532352222003058,
+      "loss": 1.5461,
+      "step": 3795
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001352911372437835,
+      "loss": 1.2607,
+      "step": 3796
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013525874803913294,
+      "loss": 1.1236,
+      "step": 3797
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001352263546099596,
+      "loss": 1.6053,
+      "step": 3798
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013519395696014476,
+      "loss": 1.6196,
+      "step": 3799
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013516155509357006,
+      "loss": 1.1496,
+      "step": 3800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013512914901411775,
+      "loss": 1.2499,
+      "step": 3801
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013509673872567058,
+      "loss": 1.0322,
+      "step": 3802
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013506432423211181,
+      "loss": 1.3462,
+      "step": 3803
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013503190553732514,
+      "loss": 1.3458,
+      "step": 3804
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001349994826451948,
+      "loss": 1.4679,
+      "step": 3805
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013496705555960558,
+      "loss": 1.5308,
+      "step": 3806
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001349346242844427,
+      "loss": 1.6142,
+      "step": 3807
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013490218882359188,
+      "loss": 1.4585,
+      "step": 3808
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013486974918093943,
+      "loss": 1.701,
+      "step": 3809
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013483730536037206,
+      "loss": 1.2185,
+      "step": 3810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013480485736577707,
+      "loss": 1.3277,
+      "step": 3811
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013477240520104212,
+      "loss": 1.4098,
+      "step": 3812
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013473994887005557,
+      "loss": 1.4008,
+      "step": 3813
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001347074883767061,
+      "loss": 1.6327,
+      "step": 3814
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013467502372488301,
+      "loss": 1.5136,
+      "step": 3815
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013464255491847603,
+      "loss": 1.2798,
+      "step": 3816
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013461008196137536,
+      "loss": 1.1132,
+      "step": 3817
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001345776048574718,
+      "loss": 1.337,
+      "step": 3818
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001345451236106566,
+      "loss": 1.5925,
+      "step": 3819
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001345126382248214,
+      "loss": 1.2878,
+      "step": 3820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013448014870385855,
+      "loss": 1.33,
+      "step": 3821
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013444765505166072,
+      "loss": 1.3551,
+      "step": 3822
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013441515727212117,
+      "loss": 1.1444,
+      "step": 3823
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013438265536913356,
+      "loss": 1.4025,
+      "step": 3824
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013435014934659213,
+      "loss": 1.2566,
+      "step": 3825
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013431763920839163,
+      "loss": 1.3581,
+      "step": 3826
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013428512495842717,
+      "loss": 1.4757,
+      "step": 3827
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001342526066005945,
+      "loss": 1.4701,
+      "step": 3828
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013422008413878982,
+      "loss": 1.3256,
+      "step": 3829
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013418755757690974,
+      "loss": 1.4374,
+      "step": 3830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013415502691885146,
+      "loss": 1.392,
+      "step": 3831
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013412249216851265,
+      "loss": 1.2186,
+      "step": 3832
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013408995332979146,
+      "loss": 1.2708,
+      "step": 3833
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013405741040658652,
+      "loss": 1.4186,
+      "step": 3834
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001340248634027969,
+      "loss": 1.3495,
+      "step": 3835
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001339923123223223,
+      "loss": 1.5036,
+      "step": 3836
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013395975716906278,
+      "loss": 1.4632,
+      "step": 3837
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013392719794691894,
+      "loss": 1.4312,
+      "step": 3838
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013389463465979186,
+      "loss": 1.4047,
+      "step": 3839
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013386206731158312,
+      "loss": 1.5614,
+      "step": 3840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013382949590619477,
+      "loss": 1.6999,
+      "step": 3841
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001337969204475293,
+      "loss": 1.4218,
+      "step": 3842
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001337643409394898,
+      "loss": 1.5219,
+      "step": 3843
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013373175738597975,
+      "loss": 1.4516,
+      "step": 3844
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013369916979090313,
+      "loss": 1.4348,
+      "step": 3845
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013366657815816447,
+      "loss": 1.5283,
+      "step": 3846
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013363398249166867,
+      "loss": 1.4681,
+      "step": 3847
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013360138279532122,
+      "loss": 1.5892,
+      "step": 3848
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013356877907302802,
+      "loss": 1.1882,
+      "step": 3849
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013353617132869552,
+      "loss": 1.2835,
+      "step": 3850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013350355956623058,
+      "loss": 1.0444,
+      "step": 3851
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013347094378954056,
+      "loss": 1.2361,
+      "step": 3852
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013343832400253337,
+      "loss": 0.9853,
+      "step": 3853
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013340570020911732,
+      "loss": 1.2246,
+      "step": 3854
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001333730724132012,
+      "loss": 1.3598,
+      "step": 3855
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013334044061869433,
+      "loss": 1.6573,
+      "step": 3856
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013330780482950647,
+      "loss": 1.3903,
+      "step": 3857
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001332751650495479,
+      "loss": 1.4641,
+      "step": 3858
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013324252128272933,
+      "loss": 1.6045,
+      "step": 3859
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013320987353296197,
+      "loss": 1.2089,
+      "step": 3860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001331772218041575,
+      "loss": 1.6887,
+      "step": 3861
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001331445661002281,
+      "loss": 1.3318,
+      "step": 3862
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013311190642508643,
+      "loss": 1.0462,
+      "step": 3863
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013307924278264555,
+      "loss": 1.1587,
+      "step": 3864
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013304657517681912,
+      "loss": 1.0649,
+      "step": 3865
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013301390361152113,
+      "loss": 1.4422,
+      "step": 3866
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013298122809066615,
+      "loss": 1.2474,
+      "step": 3867
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013294854861816918,
+      "loss": 1.2272,
+      "step": 3868
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013291586519794576,
+      "loss": 1.325,
+      "step": 3869
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013288317783391183,
+      "loss": 1.2876,
+      "step": 3870
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013285048652998378,
+      "loss": 1.098,
+      "step": 3871
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013281779129007852,
+      "loss": 1.2595,
+      "step": 3872
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001327850921181135,
+      "loss": 1.5292,
+      "step": 3873
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013275238901800645,
+      "loss": 1.1,
+      "step": 3874
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001327196819936758,
+      "loss": 1.8245,
+      "step": 3875
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001326869710490403,
+      "loss": 1.6725,
+      "step": 3876
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013265425618801918,
+      "loss": 1.8084,
+      "step": 3877
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001326215374145322,
+      "loss": 1.4706,
+      "step": 3878
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013258881473249954,
+      "loss": 1.4514,
+      "step": 3879
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013255608814584187,
+      "loss": 1.4482,
+      "step": 3880
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001325233576584803,
+      "loss": 1.2928,
+      "step": 3881
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013249062327433647,
+      "loss": 1.7002,
+      "step": 3882
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013245788499733244,
+      "loss": 1.23,
+      "step": 3883
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013242514283139068,
+      "loss": 1.5767,
+      "step": 3884
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013239239678043428,
+      "loss": 1.2567,
+      "step": 3885
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013235964684838667,
+      "loss": 1.8158,
+      "step": 3886
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001323268930391717,
+      "loss": 1.2255,
+      "step": 3887
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001322941353567139,
+      "loss": 1.2828,
+      "step": 3888
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001322613738049381,
+      "loss": 1.2923,
+      "step": 3889
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013222860838776952,
+      "loss": 1.4117,
+      "step": 3890
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000132195839109134,
+      "loss": 1.1961,
+      "step": 3891
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001321630659729578,
+      "loss": 1.4274,
+      "step": 3892
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001321302889831677,
+      "loss": 1.5434,
+      "step": 3893
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013209750814369072,
+      "loss": 1.4539,
+      "step": 3894
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013206472345845456,
+      "loss": 1.3151,
+      "step": 3895
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001320319349313874,
+      "loss": 1.2308,
+      "step": 3896
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013199914256641765,
+      "loss": 1.4787,
+      "step": 3897
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001319663463674744,
+      "loss": 1.4455,
+      "step": 3898
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001319335463384871,
+      "loss": 1.3249,
+      "step": 3899
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013190074248338576,
+      "loss": 1.429,
+      "step": 3900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013186793480610064,
+      "loss": 1.5199,
+      "step": 3901
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013183512331056268,
+      "loss": 1.6846,
+      "step": 3902
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013180230800070316,
+      "loss": 1.3475,
+      "step": 3903
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013176948888045378,
+      "loss": 1.1934,
+      "step": 3904
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001317366659537469,
+      "loss": 1.2994,
+      "step": 3905
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013170383922451505,
+      "loss": 1.6096,
+      "step": 3906
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013167100869669149,
+      "loss": 1.0494,
+      "step": 3907
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013163817437420968,
+      "loss": 1.0566,
+      "step": 3908
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013160533626100374,
+      "loss": 1.5998,
+      "step": 3909
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013157249436100818,
+      "loss": 1.3303,
+      "step": 3910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001315396486781579,
+      "loss": 1.6416,
+      "step": 3911
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013150679921638826,
+      "loss": 1.3546,
+      "step": 3912
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013147394597963526,
+      "loss": 1.1863,
+      "step": 3913
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013144108897183504,
+      "loss": 1.2233,
+      "step": 3914
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013140822819692448,
+      "loss": 1.316,
+      "step": 3915
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001313753636588407,
+      "loss": 1.344,
+      "step": 3916
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013134249536152145,
+      "loss": 1.1233,
+      "step": 3917
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001313096233089048,
+      "loss": 1.3633,
+      "step": 3918
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013127674750492926,
+      "loss": 1.421,
+      "step": 3919
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001312438679535339,
+      "loss": 1.5416,
+      "step": 3920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001312109846586582,
+      "loss": 1.1957,
+      "step": 3921
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013117809762424202,
+      "loss": 1.5797,
+      "step": 3922
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013114520685422574,
+      "loss": 1.2588,
+      "step": 3923
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013111231235255017,
+      "loss": 1.6849,
+      "step": 3924
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001310794141231565,
+      "loss": 1.1992,
+      "step": 3925
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013104651216998655,
+      "loss": 1.4897,
+      "step": 3926
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013101360649698233,
+      "loss": 1.4275,
+      "step": 3927
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001309806971080865,
+      "loss": 1.2228,
+      "step": 3928
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013094778400724207,
+      "loss": 1.5924,
+      "step": 3929
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013091486719839256,
+      "loss": 1.4531,
+      "step": 3930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013088194668548186,
+      "loss": 1.325,
+      "step": 3931
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001308490224724543,
+      "loss": 1.4039,
+      "step": 3932
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001308160945632548,
+      "loss": 1.5587,
+      "step": 3933
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013078316296182853,
+      "loss": 1.5029,
+      "step": 3934
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013075022767212118,
+      "loss": 1.5694,
+      "step": 3935
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001307172886980789,
+      "loss": 1.1516,
+      "step": 3936
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013068434604364832,
+      "loss": 1.4256,
+      "step": 3937
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001306513997127764,
+      "loss": 1.3905,
+      "step": 3938
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013061844970941062,
+      "loss": 1.1631,
+      "step": 3939
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001305854960374989,
+      "loss": 1.2035,
+      "step": 3940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013055253870098953,
+      "loss": 1.236,
+      "step": 3941
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001305195777038313,
+      "loss": 1.6956,
+      "step": 3942
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013048661304997355,
+      "loss": 1.2628,
+      "step": 3943
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013045364474336573,
+      "loss": 1.2336,
+      "step": 3944
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001304206727879581,
+      "loss": 1.4347,
+      "step": 3945
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001303876971877011,
+      "loss": 1.1341,
+      "step": 3946
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013035471794654578,
+      "loss": 1.367,
+      "step": 3947
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013032173506844343,
+      "loss": 1.3788,
+      "step": 3948
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000130288748557346,
+      "loss": 1.2321,
+      "step": 3949
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013025575841720568,
+      "loss": 1.4036,
+      "step": 3950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013022276465197522,
+      "loss": 1.1049,
+      "step": 3951
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013018976726560778,
+      "loss": 1.3653,
+      "step": 3952
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013015676626205688,
+      "loss": 1.3647,
+      "step": 3953
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013012376164527663,
+      "loss": 1.1743,
+      "step": 3954
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013009075341922138,
+      "loss": 1.3905,
+      "step": 3955
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013005774158784603,
+      "loss": 1.2705,
+      "step": 3956
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001300247261551059,
+      "loss": 1.3004,
+      "step": 3957
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012999170712495674,
+      "loss": 1.2898,
+      "step": 3958
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012995868450135472,
+      "loss": 1.1483,
+      "step": 3959
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012992565828825645,
+      "loss": 1.1869,
+      "step": 3960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001298926284896189,
+      "loss": 1.3353,
+      "step": 3961
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001298595951093996,
+      "loss": 1.044,
+      "step": 3962
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012982655815155645,
+      "loss": 0.8593,
+      "step": 3963
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001297935176200477,
+      "loss": 1.2071,
+      "step": 3964
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012976047351883211,
+      "loss": 1.4937,
+      "step": 3965
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001297274258518689,
+      "loss": 0.8907,
+      "step": 3966
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012969437462311765,
+      "loss": 1.251,
+      "step": 3967
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001296613198365384,
+      "loss": 1.2778,
+      "step": 3968
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012962826149609158,
+      "loss": 1.6405,
+      "step": 3969
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012959519960573812,
+      "loss": 1.5412,
+      "step": 3970
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012956213416943926,
+      "loss": 1.4553,
+      "step": 3971
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012952906519115678,
+      "loss": 1.5221,
+      "step": 3972
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001294959926748528,
+      "loss": 1.4335,
+      "step": 3973
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012946291662448996,
+      "loss": 1.0024,
+      "step": 3974
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012942983704403117,
+      "loss": 1.5285,
+      "step": 3975
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012939675393743995,
+      "loss": 1.5655,
+      "step": 3976
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012936366730868012,
+      "loss": 1.4963,
+      "step": 3977
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001293305771617159,
+      "loss": 1.3252,
+      "step": 3978
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012929748350051204,
+      "loss": 1.298,
+      "step": 3979
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012926438632903362,
+      "loss": 1.5941,
+      "step": 3980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012923128565124617,
+      "loss": 1.2775,
+      "step": 3981
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001291981814711157,
+      "loss": 1.2808,
+      "step": 3982
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001291650737926085,
+      "loss": 1.8077,
+      "step": 3983
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012913196261969147,
+      "loss": 1.431,
+      "step": 3984
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012909884795633171,
+      "loss": 1.3596,
+      "step": 3985
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001290657298064969,
+      "loss": 1.4112,
+      "step": 3986
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012903260817415511,
+      "loss": 1.3393,
+      "step": 3987
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012899948306327477,
+      "loss": 0.8951,
+      "step": 3988
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012896635447782475,
+      "loss": 1.2463,
+      "step": 3989
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012893322242177438,
+      "loss": 1.3332,
+      "step": 3990
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012890008689909338,
+      "loss": 1.3266,
+      "step": 3991
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012886694791375183,
+      "loss": 1.2687,
+      "step": 3992
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012883380546972032,
+      "loss": 1.2952,
+      "step": 3993
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001288006595709698,
+      "loss": 1.1347,
+      "step": 3994
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001287675102214716,
+      "loss": 1.5028,
+      "step": 3995
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012873435742519757,
+      "loss": 1.3638,
+      "step": 3996
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001287012011861199,
+      "loss": 1.229,
+      "step": 3997
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012866804150821114,
+      "loss": 1.5734,
+      "step": 3998
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012863487839544436,
+      "loss": 1.1489,
+      "step": 3999
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000128601711851793,
+      "loss": 1.2463,
+      "step": 4000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012856854188123088,
+      "loss": 1.3506,
+      "step": 4001
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012853536848773226,
+      "loss": 1.4383,
+      "step": 4002
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012850219167527186,
+      "loss": 1.4785,
+      "step": 4003
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012846901144782463,
+      "loss": 1.2266,
+      "step": 4004
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001284358278093662,
+      "loss": 1.1939,
+      "step": 4005
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012840264076387236,
+      "loss": 1.6211,
+      "step": 4006
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001283694503153195,
+      "loss": 1.5536,
+      "step": 4007
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012833625646768425,
+      "loss": 1.2907,
+      "step": 4008
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001283030592249437,
+      "loss": 1.3234,
+      "step": 4009
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012826985859107554,
+      "loss": 1.2164,
+      "step": 4010
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012823665457005752,
+      "loss": 1.413,
+      "step": 4011
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012820344716586805,
+      "loss": 1.4838,
+      "step": 4012
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001281702363824859,
+      "loss": 1.5608,
+      "step": 4013
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012813702222389016,
+      "loss": 1.3148,
+      "step": 4014
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012810380469406043,
+      "loss": 1.5144,
+      "step": 4015
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001280705837969766,
+      "loss": 1.3584,
+      "step": 4016
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001280373595366191,
+      "loss": 1.0409,
+      "step": 4017
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012800413191696864,
+      "loss": 1.476,
+      "step": 4018
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012797090094200642,
+      "loss": 1.1975,
+      "step": 4019
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012793766661571397,
+      "loss": 1.5043,
+      "step": 4020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001279044289420733,
+      "loss": 1.7085,
+      "step": 4021
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012787118792506673,
+      "loss": 1.1415,
+      "step": 4022
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012783794356867705,
+      "loss": 1.1984,
+      "step": 4023
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012780469587688744,
+      "loss": 1.4097,
+      "step": 4024
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012777144485368145,
+      "loss": 1.4991,
+      "step": 4025
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012773819050304304,
+      "loss": 1.311,
+      "step": 4026
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012770493282895663,
+      "loss": 1.5254,
+      "step": 4027
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012767167183540688,
+      "loss": 1.4617,
+      "step": 4028
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012763840752637903,
+      "loss": 1.3305,
+      "step": 4029
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012760513990585866,
+      "loss": 1.0045,
+      "step": 4030
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012757186897783168,
+      "loss": 1.4356,
+      "step": 4031
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012753859474628446,
+      "loss": 1.24,
+      "step": 4032
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012750531721520376,
+      "loss": 1.2944,
+      "step": 4033
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012747203638857667,
+      "loss": 1.5474,
+      "step": 4034
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001274387522703908,
+      "loss": 1.4412,
+      "step": 4035
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012740546486463404,
+      "loss": 1.4301,
+      "step": 4036
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001273721741752947,
+      "loss": 1.5894,
+      "step": 4037
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012733888020636153,
+      "loss": 1.3044,
+      "step": 4038
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012730558296182365,
+      "loss": 1.34,
+      "step": 4039
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012727228244567055,
+      "loss": 1.576,
+      "step": 4040
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012723897866189214,
+      "loss": 1.318,
+      "step": 4041
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012720567161447868,
+      "loss": 1.4694,
+      "step": 4042
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012717236130742085,
+      "loss": 1.3638,
+      "step": 4043
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001271390477447098,
+      "loss": 1.3756,
+      "step": 4044
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001271057309303369,
+      "loss": 1.0839,
+      "step": 4045
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000127072410868294,
+      "loss": 1.6097,
+      "step": 4046
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012703908756257342,
+      "loss": 1.3704,
+      "step": 4047
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001270057610171677,
+      "loss": 1.0478,
+      "step": 4048
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001269724312360699,
+      "loss": 1.7808,
+      "step": 4049
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012693909822327344,
+      "loss": 1.0099,
+      "step": 4050
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012690576198277206,
+      "loss": 1.2832,
+      "step": 4051
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012687242251855998,
+      "loss": 1.3638,
+      "step": 4052
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012683907983463173,
+      "loss": 1.1907,
+      "step": 4053
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012680573393498233,
+      "loss": 1.1797,
+      "step": 4054
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000126772384823607,
+      "loss": 1.199,
+      "step": 4055
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012673903250450157,
+      "loss": 1.6796,
+      "step": 4056
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001267056769816621,
+      "loss": 1.2425,
+      "step": 4057
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012667231825908506,
+      "loss": 1.2296,
+      "step": 4058
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012663895634076734,
+      "loss": 1.306,
+      "step": 4059
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001266055912307062,
+      "loss": 1.4506,
+      "step": 4060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012657222293289925,
+      "loss": 1.7092,
+      "step": 4061
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012653885145134453,
+      "loss": 1.1452,
+      "step": 4062
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012650547679004048,
+      "loss": 1.0409,
+      "step": 4063
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012647209895298578,
+      "loss": 1.3568,
+      "step": 4064
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001264387179441797,
+      "loss": 1.333,
+      "step": 4065
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001264053337676217,
+      "loss": 1.4215,
+      "step": 4066
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012637194642731175,
+      "loss": 1.1138,
+      "step": 4067
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012633855592725015,
+      "loss": 1.0808,
+      "step": 4068
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012630516227143753,
+      "loss": 1.3906,
+      "step": 4069
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012627176546387502,
+      "loss": 1.3216,
+      "step": 4070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000126238365508564,
+      "loss": 1.1267,
+      "step": 4071
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012620496240950627,
+      "loss": 1.3452,
+      "step": 4072
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012617155617070404,
+      "loss": 1.2209,
+      "step": 4073
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012613814679615995,
+      "loss": 1.1921,
+      "step": 4074
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001261047342898768,
+      "loss": 1.3209,
+      "step": 4075
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012607131865585797,
+      "loss": 1.4278,
+      "step": 4076
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012603789989810722,
+      "loss": 1.037,
+      "step": 4077
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001260044780206285,
+      "loss": 1.4883,
+      "step": 4078
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001259710530274263,
+      "loss": 1.3434,
+      "step": 4079
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012593762492250544,
+      "loss": 1.1936,
+      "step": 4080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001259041937098711,
+      "loss": 1.2485,
+      "step": 4081
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012587075939352884,
+      "loss": 1.6711,
+      "step": 4082
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012583732197748456,
+      "loss": 1.468,
+      "step": 4083
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001258038814657446,
+      "loss": 1.1889,
+      "step": 4084
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001257704378623156,
+      "loss": 1.4799,
+      "step": 4085
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001257369911712046,
+      "loss": 1.1325,
+      "step": 4086
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012570354139641906,
+      "loss": 1.3032,
+      "step": 4087
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012567008854196668,
+      "loss": 1.446,
+      "step": 4088
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012563663261185566,
+      "loss": 1.4115,
+      "step": 4089
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012560317361009452,
+      "loss": 1.6973,
+      "step": 4090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012556971154069216,
+      "loss": 1.6024,
+      "step": 4091
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001255362464076578,
+      "loss": 1.3265,
+      "step": 4092
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012550277821500106,
+      "loss": 1.226,
+      "step": 4093
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012546930696673197,
+      "loss": 1.0643,
+      "step": 4094
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012543583266686083,
+      "loss": 1.3151,
+      "step": 4095
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012540235531939838,
+      "loss": 1.5391,
+      "step": 4096
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012536887492835573,
+      "loss": 1.0429,
+      "step": 4097
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001253353914977443,
+      "loss": 1.4877,
+      "step": 4098
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001253019050315759,
+      "loss": 1.2716,
+      "step": 4099
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012526841553386274,
+      "loss": 1.4787,
+      "step": 4100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012523492300861732,
+      "loss": 1.3255,
+      "step": 4101
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012520142745985256,
+      "loss": 1.4029,
+      "step": 4102
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012516792889158174,
+      "loss": 1.2993,
+      "step": 4103
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012513442730781847,
+      "loss": 1.1391,
+      "step": 4104
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012510092271257673,
+      "loss": 1.5804,
+      "step": 4105
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012506741510987093,
+      "loss": 1.5613,
+      "step": 4106
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001250339045037157,
+      "loss": 1.4989,
+      "step": 4107
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012500039089812615,
+      "loss": 1.2509,
+      "step": 4108
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012496687429711768,
+      "loss": 1.3671,
+      "step": 4109
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001249333547047061,
+      "loss": 1.6622,
+      "step": 4110
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012489983212490754,
+      "loss": 1.0561,
+      "step": 4111
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012486630656173854,
+      "loss": 1.3313,
+      "step": 4112
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001248327780192159,
+      "loss": 1.3395,
+      "step": 4113
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012479924650135696,
+      "loss": 1.4822,
+      "step": 4114
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012476571201217913,
+      "loss": 1.4778,
+      "step": 4115
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012473217455570044,
+      "loss": 1.6572,
+      "step": 4116
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001246986341359392,
+      "loss": 1.3468,
+      "step": 4117
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012466509075691396,
+      "loss": 1.419,
+      "step": 4118
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001246315444226438,
+      "loss": 1.0718,
+      "step": 4119
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000124597995137148,
+      "loss": 1.2267,
+      "step": 4120
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012456444290444632,
+      "loss": 1.6101,
+      "step": 4121
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001245308877285588,
+      "loss": 1.2342,
+      "step": 4122
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012449732961350584,
+      "loss": 1.3352,
+      "step": 4123
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012446376856330824,
+      "loss": 1.0411,
+      "step": 4124
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012443020458198704,
+      "loss": 1.2021,
+      "step": 4125
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012439663767356375,
+      "loss": 1.4266,
+      "step": 4126
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001243630678420602,
+      "loss": 0.952,
+      "step": 4127
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012432949509149854,
+      "loss": 1.4641,
+      "step": 4128
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012429591942590126,
+      "loss": 1.4813,
+      "step": 4129
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012426234084929124,
+      "loss": 1.4853,
+      "step": 4130
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012422875936569174,
+      "loss": 1.1972,
+      "step": 4131
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012419517497912623,
+      "loss": 1.1369,
+      "step": 4132
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001241615876936187,
+      "loss": 1.1584,
+      "step": 4133
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012412799751319338,
+      "loss": 1.2224,
+      "step": 4134
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012409440444187487,
+      "loss": 1.5943,
+      "step": 4135
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012406080848368807,
+      "loss": 1.1947,
+      "step": 4136
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012402720964265837,
+      "loss": 1.3327,
+      "step": 4137
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012399360792281135,
+      "loss": 1.372,
+      "step": 4138
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012396000332817302,
+      "loss": 1.4098,
+      "step": 4139
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012392639586276973,
+      "loss": 1.1784,
+      "step": 4140
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001238927855306281,
+      "loss": 1.5904,
+      "step": 4141
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001238591723357752,
+      "loss": 1.2571,
+      "step": 4142
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012382555628223833,
+      "loss": 1.4972,
+      "step": 4143
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012379193737404527,
+      "loss": 1.2848,
+      "step": 4144
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000123758315615224,
+      "loss": 1.3199,
+      "step": 4145
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012372469100980292,
+      "loss": 1.4,
+      "step": 4146
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001236910635618108,
+      "loss": 1.3018,
+      "step": 4147
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012365743327527669,
+      "loss": 1.324,
+      "step": 4148
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012362380015422993,
+      "loss": 1.3572,
+      "step": 4149
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012359016420270037,
+      "loss": 1.1578,
+      "step": 4150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012355652542471806,
+      "loss": 1.3762,
+      "step": 4151
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012352288382431342,
+      "loss": 1.4294,
+      "step": 4152
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012348923940551723,
+      "loss": 1.2872,
+      "step": 4153
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012345559217236056,
+      "loss": 1.5771,
+      "step": 4154
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012342194212887486,
+      "loss": 1.1354,
+      "step": 4155
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001233882892790919,
+      "loss": 1.5364,
+      "step": 4156
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012335463362704382,
+      "loss": 1.1368,
+      "step": 4157
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012332097517676308,
+      "loss": 1.5627,
+      "step": 4158
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001232873139322824,
+      "loss": 1.2049,
+      "step": 4159
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012325364989763496,
+      "loss": 1.6257,
+      "step": 4160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012321998307685418,
+      "loss": 1.6416,
+      "step": 4161
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012318631347397384,
+      "loss": 1.3545,
+      "step": 4162
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012315264109302807,
+      "loss": 1.2765,
+      "step": 4163
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012311896593805136,
+      "loss": 1.0102,
+      "step": 4164
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001230852880130784,
+      "loss": 1.1928,
+      "step": 4165
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012305160732214442,
+      "loss": 1.0357,
+      "step": 4166
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001230179238692848,
+      "loss": 1.6036,
+      "step": 4167
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012298423765853532,
+      "loss": 1.3121,
+      "step": 4168
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012295054869393212,
+      "loss": 1.0338,
+      "step": 4169
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012291685697951162,
+      "loss": 0.9909,
+      "step": 4170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012288316251931057,
+      "loss": 0.8606,
+      "step": 4171
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001228494653173661,
+      "loss": 1.3049,
+      "step": 4172
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012281576537771563,
+      "loss": 1.1885,
+      "step": 4173
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012278206270439692,
+      "loss": 1.5874,
+      "step": 4174
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012274835730144803,
+      "loss": 1.3904,
+      "step": 4175
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012271464917290736,
+      "loss": 1.2702,
+      "step": 4176
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012268093832281372,
+      "loss": 1.1879,
+      "step": 4177
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012264722475520607,
+      "loss": 1.3079,
+      "step": 4178
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012261350847412382,
+      "loss": 1.4645,
+      "step": 4179
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012257978948360676,
+      "loss": 1.3188,
+      "step": 4180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012254606778769485,
+      "loss": 1.2044,
+      "step": 4181
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001225123433904285,
+      "loss": 1.0778,
+      "step": 4182
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012247861629584834,
+      "loss": 1.5433,
+      "step": 4183
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012244488650799544,
+      "loss": 1.1533,
+      "step": 4184
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012241115403091105,
+      "loss": 1.543,
+      "step": 4185
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001223774188686369,
+      "loss": 1.5763,
+      "step": 4186
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012234368102521492,
+      "loss": 1.2167,
+      "step": 4187
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012230994050468745,
+      "loss": 1.2447,
+      "step": 4188
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012227619731109706,
+      "loss": 1.283,
+      "step": 4189
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001222424514484867,
+      "loss": 1.6784,
+      "step": 4190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012220870292089963,
+      "loss": 0.9396,
+      "step": 4191
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012217495173237942,
+      "loss": 1.3131,
+      "step": 4192
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012214119788696998,
+      "loss": 1.6188,
+      "step": 4193
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012210744138871552,
+      "loss": 1.1168,
+      "step": 4194
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012207368224166057,
+      "loss": 1.352,
+      "step": 4195
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012203992044984995,
+      "loss": 1.2233,
+      "step": 4196
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012200615601732885,
+      "loss": 1.6411,
+      "step": 4197
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012197238894814278,
+      "loss": 1.6955,
+      "step": 4198
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012193861924633747,
+      "loss": 1.3146,
+      "step": 4199
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012190484691595909,
+      "loss": 1.6051,
+      "step": 4200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012187107196105407,
+      "loss": 1.5888,
+      "step": 4201
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012183729438566909,
+      "loss": 1.265,
+      "step": 4202
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012180351419385125,
+      "loss": 1.6637,
+      "step": 4203
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012176973138964793,
+      "loss": 1.2273,
+      "step": 4204
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001217359459771068,
+      "loss": 1.4264,
+      "step": 4205
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012170215796027586,
+      "loss": 1.3015,
+      "step": 4206
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001216683673432034,
+      "loss": 1.6712,
+      "step": 4207
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012163457412993804,
+      "loss": 1.4582,
+      "step": 4208
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012160077832452872,
+      "loss": 1.4348,
+      "step": 4209
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001215669799310247,
+      "loss": 1.4729,
+      "step": 4210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012153317895347551,
+      "loss": 1.2496,
+      "step": 4211
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000121499375395931,
+      "loss": 1.3514,
+      "step": 4212
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012146556926244135,
+      "loss": 1.1948,
+      "step": 4213
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012143176055705707,
+      "loss": 1.3097,
+      "step": 4214
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012139794928382889,
+      "loss": 1.5947,
+      "step": 4215
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012136413544680792,
+      "loss": 1.4042,
+      "step": 4216
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012133031905004554,
+      "loss": 1.1791,
+      "step": 4217
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012129650009759354,
+      "loss": 1.2949,
+      "step": 4218
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012126267859350386,
+      "loss": 1.0607,
+      "step": 4219
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012122885454182884,
+      "loss": 1.5365,
+      "step": 4220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012119502794662114,
+      "loss": 1.1645,
+      "step": 4221
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012116119881193362,
+      "loss": 1.3007,
+      "step": 4222
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012112736714181953,
+      "loss": 1.1749,
+      "step": 4223
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012109353294033248,
+      "loss": 1.3585,
+      "step": 4224
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001210596962115262,
+      "loss": 1.3516,
+      "step": 4225
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012102585695945492,
+      "loss": 1.1991,
+      "step": 4226
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012099201518817305,
+      "loss": 1.4106,
+      "step": 4227
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012095817090173535,
+      "loss": 1.3676,
+      "step": 4228
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012092432410419685,
+      "loss": 1.1199,
+      "step": 4229
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001208904747996129,
+      "loss": 1.7095,
+      "step": 4230
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012085662299203922,
+      "loss": 1.1405,
+      "step": 4231
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012082276868553167,
+      "loss": 1.4541,
+      "step": 4232
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012078891188414652,
+      "loss": 1.45,
+      "step": 4233
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012075505259194033,
+      "loss": 1.3659,
+      "step": 4234
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012072119081297,
+      "loss": 1.1551,
+      "step": 4235
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012068732655129258,
+      "loss": 1.4984,
+      "step": 4236
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012065345981096556,
+      "loss": 1.471,
+      "step": 4237
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012061959059604671,
+      "loss": 1.5304,
+      "step": 4238
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012058571891059398,
+      "loss": 1.3408,
+      "step": 4239
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012055184475866579,
+      "loss": 1.4449,
+      "step": 4240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012051796814432075,
+      "loss": 1.5665,
+      "step": 4241
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012048408907161772,
+      "loss": 1.1174,
+      "step": 4242
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00012045020754461597,
+      "loss": 1.2486,
+      "step": 4243
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00012041632356737502,
+      "loss": 1.6525,
+      "step": 4244
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00012038243714395465,
+      "loss": 1.244,
+      "step": 4245
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00012034854827841497,
+      "loss": 1.5098,
+      "step": 4246
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00012031465697481636,
+      "loss": 1.2791,
+      "step": 4247
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00012028076323721953,
+      "loss": 1.2261,
+      "step": 4248
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00012024686706968542,
+      "loss": 1.4783,
+      "step": 4249
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001202129684762753,
+      "loss": 1.5022,
+      "step": 4250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00012017906746105076,
+      "loss": 1.1444,
+      "step": 4251
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001201451640280736,
+      "loss": 1.2819,
+      "step": 4252
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00012011125818140599,
+      "loss": 1.1467,
+      "step": 4253
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00012007734992511033,
+      "loss": 1.5221,
+      "step": 4254
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00012004343926324935,
+      "loss": 0.9617,
+      "step": 4255
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00012000952619988604,
+      "loss": 1.2124,
+      "step": 4256
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001199756107390837,
+      "loss": 1.6676,
+      "step": 4257
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011994169288490591,
+      "loss": 1.6159,
+      "step": 4258
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011990777264141653,
+      "loss": 1.3897,
+      "step": 4259
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001198738500126797,
+      "loss": 1.3691,
+      "step": 4260
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011983992500275987,
+      "loss": 1.4661,
+      "step": 4261
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011980599761572176,
+      "loss": 1.5879,
+      "step": 4262
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011977206785563033,
+      "loss": 1.4504,
+      "step": 4263
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011973813572655095,
+      "loss": 1.1042,
+      "step": 4264
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011970420123254916,
+      "loss": 1.3027,
+      "step": 4265
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001196702643776908,
+      "loss": 1.406,
+      "step": 4266
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011963632516604199,
+      "loss": 1.3022,
+      "step": 4267
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011960238360166924,
+      "loss": 1.5079,
+      "step": 4268
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011956843968863918,
+      "loss": 1.4669,
+      "step": 4269
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011953449343101879,
+      "loss": 1.1678,
+      "step": 4270
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011950054483287541,
+      "loss": 1.1237,
+      "step": 4271
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001194665938982765,
+      "loss": 1.2211,
+      "step": 4272
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011943264063128994,
+      "loss": 1.2217,
+      "step": 4273
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011939868503598381,
+      "loss": 1.1954,
+      "step": 4274
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011936472711642655,
+      "loss": 1.5292,
+      "step": 4275
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011933076687668674,
+      "loss": 1.396,
+      "step": 4276
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011929680432083338,
+      "loss": 1.4956,
+      "step": 4277
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011926283945293568,
+      "loss": 1.337,
+      "step": 4278
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011922887227706314,
+      "loss": 1.3088,
+      "step": 4279
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001191949027972855,
+      "loss": 1.4343,
+      "step": 4280
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011916093101767285,
+      "loss": 1.615,
+      "step": 4281
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001191269569422955,
+      "loss": 1.3547,
+      "step": 4282
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011909298057522402,
+      "loss": 1.0658,
+      "step": 4283
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011905900192052937,
+      "loss": 1.4159,
+      "step": 4284
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011902502098228261,
+      "loss": 1.172,
+      "step": 4285
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011899103776455518,
+      "loss": 1.3406,
+      "step": 4286
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001189570522714188,
+      "loss": 1.1172,
+      "step": 4287
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011892306450694545,
+      "loss": 0.9568,
+      "step": 4288
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011888907447520734,
+      "loss": 1.2208,
+      "step": 4289
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011885508218027696,
+      "loss": 1.31,
+      "step": 4290
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011882108762622716,
+      "loss": 1.6381,
+      "step": 4291
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011878709081713093,
+      "loss": 1.4149,
+      "step": 4292
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011875309175706162,
+      "loss": 1.3633,
+      "step": 4293
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011871909045009282,
+      "loss": 1.1267,
+      "step": 4294
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011868508690029842,
+      "loss": 1.158,
+      "step": 4295
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011865108111175251,
+      "loss": 1.6291,
+      "step": 4296
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001186170730885295,
+      "loss": 1.1978,
+      "step": 4297
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011858306283470409,
+      "loss": 1.2466,
+      "step": 4298
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011854905035435116,
+      "loss": 1.3215,
+      "step": 4299
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011851503565154594,
+      "loss": 1.3964,
+      "step": 4300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001184810187303639,
+      "loss": 1.6698,
+      "step": 4301
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011844699959488077,
+      "loss": 1.5367,
+      "step": 4302
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011841297824917255,
+      "loss": 1.2111,
+      "step": 4303
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001183789546973155,
+      "loss": 1.2489,
+      "step": 4304
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011834492894338615,
+      "loss": 1.585,
+      "step": 4305
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011831090099146131,
+      "loss": 1.4296,
+      "step": 4306
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011827687084561798,
+      "loss": 1.3221,
+      "step": 4307
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011824283850993356,
+      "loss": 1.3904,
+      "step": 4308
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011820880398848557,
+      "loss": 1.0704,
+      "step": 4309
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011817476728535184,
+      "loss": 1.2461,
+      "step": 4310
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011814072840461052,
+      "loss": 1.0797,
+      "step": 4311
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011810668735033999,
+      "loss": 1.1842,
+      "step": 4312
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011807264412661879,
+      "loss": 1.4172,
+      "step": 4313
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011803859873752585,
+      "loss": 1.4509,
+      "step": 4314
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011800455118714038,
+      "loss": 1.2408,
+      "step": 4315
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011797050147954165,
+      "loss": 1.59,
+      "step": 4316
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001179364496188094,
+      "loss": 1.242,
+      "step": 4317
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001179023956090236,
+      "loss": 1.8429,
+      "step": 4318
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011786833945426434,
+      "loss": 1.6778,
+      "step": 4319
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011783428115861208,
+      "loss": 1.5854,
+      "step": 4320
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011780022072614755,
+      "loss": 1.3771,
+      "step": 4321
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011776615816095164,
+      "loss": 1.3946,
+      "step": 4322
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011773209346710556,
+      "loss": 0.9062,
+      "step": 4323
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001176980266486908,
+      "loss": 1.3225,
+      "step": 4324
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011766395770978907,
+      "loss": 1.0655,
+      "step": 4325
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011762988665448229,
+      "loss": 1.3373,
+      "step": 4326
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011759581348685271,
+      "loss": 1.3639,
+      "step": 4327
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011756173821098284,
+      "loss": 1.1085,
+      "step": 4328
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011752766083095535,
+      "loss": 1.3766,
+      "step": 4329
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001174935813508532,
+      "loss": 1.2594,
+      "step": 4330
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011745949977475969,
+      "loss": 1.3921,
+      "step": 4331
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011742541610675826,
+      "loss": 1.2915,
+      "step": 4332
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011739133035093263,
+      "loss": 1.3034,
+      "step": 4333
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011735724251136682,
+      "loss": 1.4083,
+      "step": 4334
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011732315259214503,
+      "loss": 1.1395,
+      "step": 4335
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011728906059735172,
+      "loss": 1.2654,
+      "step": 4336
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011725496653107166,
+      "loss": 1.4486,
+      "step": 4337
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011722087039738983,
+      "loss": 1.357,
+      "step": 4338
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001171867722003914,
+      "loss": 1.274,
+      "step": 4339
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011715267194416188,
+      "loss": 0.9442,
+      "step": 4340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.000117118569632787,
+      "loss": 1.2633,
+      "step": 4341
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011708446527035273,
+      "loss": 1.5633,
+      "step": 4342
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011705035886094522,
+      "loss": 1.1932,
+      "step": 4343
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011701625040865096,
+      "loss": 1.6143,
+      "step": 4344
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001169821399175567,
+      "loss": 1.3407,
+      "step": 4345
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001169480273917493,
+      "loss": 1.4359,
+      "step": 4346
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011691391283531601,
+      "loss": 1.0451,
+      "step": 4347
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011687979625234421,
+      "loss": 1.2902,
+      "step": 4348
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011684567764692165,
+      "loss": 1.539,
+      "step": 4349
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011681155702313616,
+      "loss": 1.3258,
+      "step": 4350
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011677743438507593,
+      "loss": 1.2446,
+      "step": 4351
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011674330973682938,
+      "loss": 1.5448,
+      "step": 4352
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011670918308248515,
+      "loss": 1.2294,
+      "step": 4353
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011667505442613212,
+      "loss": 1.6053,
+      "step": 4354
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001166409237718594,
+      "loss": 0.8969,
+      "step": 4355
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011660679112375632,
+      "loss": 1.7142,
+      "step": 4356
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011657265648591254,
+      "loss": 1.318,
+      "step": 4357
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011653851986241786,
+      "loss": 1.674,
+      "step": 4358
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011650438125736236,
+      "loss": 1.1144,
+      "step": 4359
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011647024067483636,
+      "loss": 1.0936,
+      "step": 4360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011643609811893044,
+      "loss": 1.8264,
+      "step": 4361
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011640195359373529,
+      "loss": 1.5969,
+      "step": 4362
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011636780710334202,
+      "loss": 1.1143,
+      "step": 4363
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011633365865184187,
+      "loss": 1.1627,
+      "step": 4364
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011629950824332634,
+      "loss": 1.5278,
+      "step": 4365
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011626535588188714,
+      "loss": 1.537,
+      "step": 4366
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001162312015716162,
+      "loss": 1.8849,
+      "step": 4367
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001161970453166058,
+      "loss": 1.3478,
+      "step": 4368
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011616288712094828,
+      "loss": 1.2766,
+      "step": 4369
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011612872698873634,
+      "loss": 1.5726,
+      "step": 4370
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011609456492406285,
+      "loss": 1.1896,
+      "step": 4371
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011606040093102099,
+      "loss": 1.5415,
+      "step": 4372
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011602623501370407,
+      "loss": 1.4345,
+      "step": 4373
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011599206717620568,
+      "loss": 1.65,
+      "step": 4374
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011595789742261964,
+      "loss": 1.0358,
+      "step": 4375
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011592372575703998,
+      "loss": 1.295,
+      "step": 4376
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011588955218356099,
+      "loss": 1.249,
+      "step": 4377
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011585537670627717,
+      "loss": 1.1792,
+      "step": 4378
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011582119932928328,
+      "loss": 1.2685,
+      "step": 4379
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011578702005667424,
+      "loss": 1.3914,
+      "step": 4380
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001157528388925452,
+      "loss": 1.2916,
+      "step": 4381
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011571865584099168,
+      "loss": 1.1325,
+      "step": 4382
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001156844709061092,
+      "loss": 1.4876,
+      "step": 4383
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001156502840919937,
+      "loss": 1.282,
+      "step": 4384
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011561609540274128,
+      "loss": 1.2367,
+      "step": 4385
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011558190484244819,
+      "loss": 1.2124,
+      "step": 4386
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011554771241521101,
+      "loss": 1.3653,
+      "step": 4387
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011551351812512648,
+      "loss": 1.2742,
+      "step": 4388
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011547932197629159,
+      "loss": 1.3765,
+      "step": 4389
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011544512397280356,
+      "loss": 1.3093,
+      "step": 4390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011541092411875985,
+      "loss": 1.336,
+      "step": 4391
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011537672241825803,
+      "loss": 1.212,
+      "step": 4392
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011534251887539603,
+      "loss": 1.2266,
+      "step": 4393
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011530831349427194,
+      "loss": 1.3114,
+      "step": 4394
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011527410627898408,
+      "loss": 1.3519,
+      "step": 4395
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011523989723363094,
+      "loss": 1.6955,
+      "step": 4396
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001152056863623113,
+      "loss": 1.3406,
+      "step": 4397
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011517147366912417,
+      "loss": 1.1648,
+      "step": 4398
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011513725915816866,
+      "loss": 1.5173,
+      "step": 4399
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011510304283354425,
+      "loss": 1.442,
+      "step": 4400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011506882469935052,
+      "loss": 1.2579,
+      "step": 4401
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011503460475968735,
+      "loss": 1.256,
+      "step": 4402
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011500038301865477,
+      "loss": 1.4007,
+      "step": 4403
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011496615948035303,
+      "loss": 1.5559,
+      "step": 4404
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011493193414888268,
+      "loss": 1.4581,
+      "step": 4405
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001148977070283444,
+      "loss": 1.4594,
+      "step": 4406
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011486347812283906,
+      "loss": 1.5512,
+      "step": 4407
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011482924743646783,
+      "loss": 1.3009,
+      "step": 4408
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011479501497333212,
+      "loss": 1.1048,
+      "step": 4409
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011476078073753339,
+      "loss": 1.1399,
+      "step": 4410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011472654473317343,
+      "loss": 1.0315,
+      "step": 4411
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011469230696435429,
+      "loss": 1.4386,
+      "step": 4412
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011465806743517809,
+      "loss": 1.412,
+      "step": 4413
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011462382614974725,
+      "loss": 1.1849,
+      "step": 4414
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011458958311216441,
+      "loss": 1.4809,
+      "step": 4415
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011455533832653236,
+      "loss": 1.6643,
+      "step": 4416
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011452109179695418,
+      "loss": 1.3232,
+      "step": 4417
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011448684352753308,
+      "loss": 1.4277,
+      "step": 4418
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011445259352237255,
+      "loss": 1.3318,
+      "step": 4419
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011441834178557617,
+      "loss": 1.4172,
+      "step": 4420
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011438408832124789,
+      "loss": 1.4666,
+      "step": 4421
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011434983313349179,
+      "loss": 1.4372,
+      "step": 4422
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011431557622641206,
+      "loss": 1.1551,
+      "step": 4423
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011428131760411324,
+      "loss": 1.4743,
+      "step": 4424
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011424705727070008,
+      "loss": 1.3254,
+      "step": 4425
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011421279523027742,
+      "loss": 1.0423,
+      "step": 4426
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011417853148695032,
+      "loss": 1.0494,
+      "step": 4427
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001141442660448242,
+      "loss": 1.0984,
+      "step": 4428
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011410999890800446,
+      "loss": 1.2389,
+      "step": 4429
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011407573008059687,
+      "loss": 1.4676,
+      "step": 4430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011404145956670731,
+      "loss": 1.738,
+      "step": 4431
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011400718737044196,
+      "loss": 1.3196,
+      "step": 4432
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011397291349590706,
+      "loss": 1.315,
+      "step": 4433
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001139386379472092,
+      "loss": 1.2986,
+      "step": 4434
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011390436072845507,
+      "loss": 1.4037,
+      "step": 4435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011387008184375156,
+      "loss": 1.12,
+      "step": 4436
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011383580129720586,
+      "loss": 1.1888,
+      "step": 4437
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011380151909292522,
+      "loss": 1.6269,
+      "step": 4438
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011376723523501723,
+      "loss": 1.2048,
+      "step": 4439
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011373294972758952,
+      "loss": 1.4629,
+      "step": 4440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011369866257475006,
+      "loss": 1.5489,
+      "step": 4441
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011366437378060699,
+      "loss": 1.5154,
+      "step": 4442
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011363008334926855,
+      "loss": 1.3408,
+      "step": 4443
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011359579128484326,
+      "loss": 1.2709,
+      "step": 4444
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011356149759143986,
+      "loss": 1.1577,
+      "step": 4445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001135272022731672,
+      "loss": 1.5552,
+      "step": 4446
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011349290533413439,
+      "loss": 1.0018,
+      "step": 4447
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011345860677845071,
+      "loss": 1.3695,
+      "step": 4448
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011342430661022566,
+      "loss": 1.2742,
+      "step": 4449
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011339000483356886,
+      "loss": 1.4131,
+      "step": 4450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001133557014525902,
+      "loss": 1.4684,
+      "step": 4451
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011332139647139976,
+      "loss": 1.0398,
+      "step": 4452
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011328708989410777,
+      "loss": 1.3393,
+      "step": 4453
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011325278172482465,
+      "loss": 1.4957,
+      "step": 4454
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011321847196766104,
+      "loss": 1.265,
+      "step": 4455
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011318416062672779,
+      "loss": 1.3647,
+      "step": 4456
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011314984770613583,
+      "loss": 1.1843,
+      "step": 4457
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011311553320999644,
+      "loss": 1.0887,
+      "step": 4458
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000113081217142421,
+      "loss": 1.5315,
+      "step": 4459
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011304689950752101,
+      "loss": 1.0296,
+      "step": 4460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011301258030940831,
+      "loss": 1.4352,
+      "step": 4461
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011297825955219486,
+      "loss": 1.3268,
+      "step": 4462
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011294393723999272,
+      "loss": 1.9027,
+      "step": 4463
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011290961337691428,
+      "loss": 1.2614,
+      "step": 4464
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011287528796707206,
+      "loss": 1.2503,
+      "step": 4465
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011284096101457869,
+      "loss": 1.3977,
+      "step": 4466
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011280663252354707,
+      "loss": 1.2166,
+      "step": 4467
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011277230249809028,
+      "loss": 1.2007,
+      "step": 4468
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011273797094232161,
+      "loss": 0.909,
+      "step": 4469
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001127036378603544,
+      "loss": 1.6466,
+      "step": 4470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011266930325630231,
+      "loss": 1.4195,
+      "step": 4471
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011263496713427919,
+      "loss": 1.3349,
+      "step": 4472
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011260062949839891,
+      "loss": 1.2752,
+      "step": 4473
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011256629035277571,
+      "loss": 1.4084,
+      "step": 4474
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011253194970152392,
+      "loss": 1.2835,
+      "step": 4475
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011249760754875802,
+      "loss": 1.2133,
+      "step": 4476
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011246326389859274,
+      "loss": 1.3562,
+      "step": 4477
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011242891875514297,
+      "loss": 1.6592,
+      "step": 4478
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011239457212252375,
+      "loss": 1.147,
+      "step": 4479
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011236022400485033,
+      "loss": 1.1446,
+      "step": 4480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011232587440623812,
+      "loss": 1.3529,
+      "step": 4481
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011229152333080272,
+      "loss": 1.2851,
+      "step": 4482
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001122571707826599,
+      "loss": 1.4623,
+      "step": 4483
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011222281676592559,
+      "loss": 1.5317,
+      "step": 4484
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011218846128471591,
+      "loss": 0.9298,
+      "step": 4485
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011215410434314721,
+      "loss": 1.5309,
+      "step": 4486
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001121197459453359,
+      "loss": 1.2726,
+      "step": 4487
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011208538609539865,
+      "loss": 1.3815,
+      "step": 4488
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011205102479745231,
+      "loss": 1.4644,
+      "step": 4489
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011201666205561383,
+      "loss": 1.3048,
+      "step": 4490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011198229787400038,
+      "loss": 1.4629,
+      "step": 4491
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011194793225672935,
+      "loss": 1.5578,
+      "step": 4492
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001119135652079182,
+      "loss": 0.9898,
+      "step": 4493
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001118791967316846,
+      "loss": 1.0708,
+      "step": 4494
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001118448268321465,
+      "loss": 1.1744,
+      "step": 4495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011181045551342182,
+      "loss": 1.228,
+      "step": 4496
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011177608277962881,
+      "loss": 1.2589,
+      "step": 4497
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011174170863488582,
+      "loss": 1.2271,
+      "step": 4498
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001117073330833114,
+      "loss": 1.4609,
+      "step": 4499
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011167295612902425,
+      "loss": 1.5124,
+      "step": 4500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011163857777614319,
+      "loss": 1.1758,
+      "step": 4501
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011160419802878736,
+      "loss": 1.3418,
+      "step": 4502
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011156981689107586,
+      "loss": 1.0841,
+      "step": 4503
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011153543436712812,
+      "loss": 1.3296,
+      "step": 4504
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011150105046106367,
+      "loss": 1.2831,
+      "step": 4505
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011146666517700222,
+      "loss": 1.0373,
+      "step": 4506
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011143227851906359,
+      "loss": 1.2499,
+      "step": 4507
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011139789049136786,
+      "loss": 1.3041,
+      "step": 4508
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011136350109803526,
+      "loss": 1.5403,
+      "step": 4509
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011132911034318608,
+      "loss": 1.3771,
+      "step": 4510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011129471823094086,
+      "loss": 1.3736,
+      "step": 4511
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011126032476542034,
+      "loss": 1.1481,
+      "step": 4512
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001112259299507453,
+      "loss": 1.0401,
+      "step": 4513
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011119153379103678,
+      "loss": 1.5737,
+      "step": 4514
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011115713629041596,
+      "loss": 1.299,
+      "step": 4515
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011112273745300416,
+      "loss": 1.4871,
+      "step": 4516
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001110883372829229,
+      "loss": 1.3863,
+      "step": 4517
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011105393578429377,
+      "loss": 1.6752,
+      "step": 4518
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011101953296123867,
+      "loss": 1.3751,
+      "step": 4519
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011098512881787951,
+      "loss": 1.4419,
+      "step": 4520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011095072335833838,
+      "loss": 1.3756,
+      "step": 4521
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001109163165867377,
+      "loss": 1.1717,
+      "step": 4522
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011088190850719975,
+      "loss": 1.5597,
+      "step": 4523
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011084749912384724,
+      "loss": 1.4844,
+      "step": 4524
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001108130884408029,
+      "loss": 1.2047,
+      "step": 4525
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011077867646218965,
+      "loss": 1.0782,
+      "step": 4526
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011074426319213052,
+      "loss": 1.3272,
+      "step": 4527
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011070984863474876,
+      "loss": 1.4124,
+      "step": 4528
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011067543279416777,
+      "loss": 1.1234,
+      "step": 4529
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011064101567451102,
+      "loss": 1.3998,
+      "step": 4530
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011060659727990223,
+      "loss": 1.1954,
+      "step": 4531
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011057217761446523,
+      "loss": 1.5785,
+      "step": 4532
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011053775668232402,
+      "loss": 1.2645,
+      "step": 4533
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011050333448760273,
+      "loss": 1.3764,
+      "step": 4534
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011046891103442565,
+      "loss": 1.0885,
+      "step": 4535
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011043448632691722,
+      "loss": 1.1168,
+      "step": 4536
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011040006036920206,
+      "loss": 1.5231,
+      "step": 4537
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011036563316540485,
+      "loss": 1.3603,
+      "step": 4538
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011033120471965054,
+      "loss": 1.3212,
+      "step": 4539
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011029677503606415,
+      "loss": 1.6239,
+      "step": 4540
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011026234411877085,
+      "loss": 1.2469,
+      "step": 4541
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011022791197189603,
+      "loss": 1.4444,
+      "step": 4542
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001101934785995651,
+      "loss": 1.5147,
+      "step": 4543
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011015904400590375,
+      "loss": 1.487,
+      "step": 4544
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011012460819503771,
+      "loss": 1.0693,
+      "step": 4545
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011009017117109298,
+      "loss": 1.1446,
+      "step": 4546
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011005573293819553,
+      "loss": 1.3046,
+      "step": 4547
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011002129350047162,
+      "loss": 1.2102,
+      "step": 4548
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010998685286204761,
+      "loss": 1.401,
+      "step": 4549
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010995241102705001,
+      "loss": 1.6104,
+      "step": 4550
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010991796799960544,
+      "loss": 1.4151,
+      "step": 4551
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010988352378384073,
+      "loss": 1.7013,
+      "step": 4552
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010984907838388272,
+      "loss": 1.3449,
+      "step": 4553
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010981463180385858,
+      "loss": 1.2791,
+      "step": 4554
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010978018404789545,
+      "loss": 1.2898,
+      "step": 4555
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010974573512012077,
+      "loss": 1.5075,
+      "step": 4556
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010971128502466192,
+      "loss": 1.5311,
+      "step": 4557
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010967683376564664,
+      "loss": 1.6528,
+      "step": 4558
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010964238134720264,
+      "loss": 1.2506,
+      "step": 4559
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010960792777345785,
+      "loss": 1.5081,
+      "step": 4560
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010957347304854031,
+      "loss": 1.6248,
+      "step": 4561
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010953901717657824,
+      "loss": 1.0439,
+      "step": 4562
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010950456016169994,
+      "loss": 1.4073,
+      "step": 4563
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010947010200803389,
+      "loss": 1.3119,
+      "step": 4564
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010943564271970866,
+      "loss": 1.2868,
+      "step": 4565
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010940118230085303,
+      "loss": 1.4188,
+      "step": 4566
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010936672075559581,
+      "loss": 1.2614,
+      "step": 4567
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010933225808806604,
+      "loss": 1.3498,
+      "step": 4568
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010929779430239291,
+      "loss": 1.2101,
+      "step": 4569
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001092633294027056,
+      "loss": 1.3193,
+      "step": 4570
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010922886339313359,
+      "loss": 1.4178,
+      "step": 4571
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010919439627780641,
+      "loss": 1.121,
+      "step": 4572
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010915992806085369,
+      "loss": 1.146,
+      "step": 4573
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010912545874640528,
+      "loss": 1.6456,
+      "step": 4574
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010909098833859108,
+      "loss": 1.3031,
+      "step": 4575
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010905651684154121,
+      "loss": 1.3854,
+      "step": 4576
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010902204425938583,
+      "loss": 1.4817,
+      "step": 4577
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010898757059625527,
+      "loss": 0.971,
+      "step": 4578
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010895309585628,
+      "loss": 1.464,
+      "step": 4579
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010891862004359058,
+      "loss": 1.2535,
+      "step": 4580
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010888414316231776,
+      "loss": 1.4263,
+      "step": 4581
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010884966521659241,
+      "loss": 1.31,
+      "step": 4582
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001088151862105454,
+      "loss": 1.4422,
+      "step": 4583
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001087807061483079,
+      "loss": 1.2381,
+      "step": 4584
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010874622503401115,
+      "loss": 1.2015,
+      "step": 4585
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010871174287178648,
+      "loss": 1.4597,
+      "step": 4586
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010867725966576532,
+      "loss": 1.4477,
+      "step": 4587
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010864277542007932,
+      "loss": 1.7601,
+      "step": 4588
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010860829013886023,
+      "loss": 1.394,
+      "step": 4589
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010857380382623986,
+      "loss": 1.6071,
+      "step": 4590
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010853931648635016,
+      "loss": 1.4794,
+      "step": 4591
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010850482812332326,
+      "loss": 1.2264,
+      "step": 4592
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001084703387412914,
+      "loss": 1.0461,
+      "step": 4593
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010843584834438689,
+      "loss": 1.353,
+      "step": 4594
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001084013569367422,
+      "loss": 1.5948,
+      "step": 4595
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010836686452248992,
+      "loss": 1.5641,
+      "step": 4596
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010833237110576276,
+      "loss": 1.1409,
+      "step": 4597
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001082978766906935,
+      "loss": 1.1146,
+      "step": 4598
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010826338128141515,
+      "loss": 1.8988,
+      "step": 4599
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010822888488206073,
+      "loss": 1.05,
+      "step": 4600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010819438749676342,
+      "loss": 1.4,
+      "step": 4601
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010815988912965653,
+      "loss": 1.3288,
+      "step": 4602
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010812538978487353,
+      "loss": 1.3611,
+      "step": 4603
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010809088946654784,
+      "loss": 1.5957,
+      "step": 4604
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001080563881788132,
+      "loss": 1.1572,
+      "step": 4605
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001080218859258034,
+      "loss": 1.5984,
+      "step": 4606
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010798738271165223,
+      "loss": 1.4051,
+      "step": 4607
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010795287854049376,
+      "loss": 1.2907,
+      "step": 4608
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010791837341646209,
+      "loss": 1.4852,
+      "step": 4609
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010788386734369143,
+      "loss": 1.4135,
+      "step": 4610
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001078493603263161,
+      "loss": 1.0676,
+      "step": 4611
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010781485236847063,
+      "loss": 1.2827,
+      "step": 4612
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010778034347428951,
+      "loss": 1.3545,
+      "step": 4613
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010774583364790747,
+      "loss": 1.5576,
+      "step": 4614
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010771132289345924,
+      "loss": 1.4628,
+      "step": 4615
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010767681121507981,
+      "loss": 1.2757,
+      "step": 4616
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010764229861690414,
+      "loss": 1.0495,
+      "step": 4617
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010760778510306733,
+      "loss": 1.3707,
+      "step": 4618
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010757327067770469,
+      "loss": 1.3877,
+      "step": 4619
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010753875534495146,
+      "loss": 1.2944,
+      "step": 4620
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010750423910894315,
+      "loss": 1.4976,
+      "step": 4621
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010746972197381532,
+      "loss": 1.2888,
+      "step": 4622
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010743520394370366,
+      "loss": 1.1724,
+      "step": 4623
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010740068502274387,
+      "loss": 1.3519,
+      "step": 4624
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001073661652150719,
+      "loss": 1.3131,
+      "step": 4625
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010733164452482371,
+      "loss": 1.084,
+      "step": 4626
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010729712295613537,
+      "loss": 1.2325,
+      "step": 4627
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001072626005131431,
+      "loss": 1.459,
+      "step": 4628
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010722807719998322,
+      "loss": 1.2982,
+      "step": 4629
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010719355302079213,
+      "loss": 0.8691,
+      "step": 4630
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010715902797970629,
+      "loss": 1.0417,
+      "step": 4631
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010712450208086237,
+      "loss": 1.0678,
+      "step": 4632
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010708997532839708,
+      "loss": 1.1581,
+      "step": 4633
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010705544772644722,
+      "loss": 1.3861,
+      "step": 4634
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010702091927914973,
+      "loss": 1.346,
+      "step": 4635
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010698638999064161,
+      "loss": 1.2338,
+      "step": 4636
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010695185986505997,
+      "loss": 1.1525,
+      "step": 4637
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010691732890654209,
+      "loss": 1.6936,
+      "step": 4638
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010688279711922524,
+      "loss": 1.3594,
+      "step": 4639
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010684826450724688,
+      "loss": 1.6059,
+      "step": 4640
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010681373107474448,
+      "loss": 1.101,
+      "step": 4641
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010677919682585573,
+      "loss": 1.0476,
+      "step": 4642
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001067446617647183,
+      "loss": 1.3285,
+      "step": 4643
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010671012589546998,
+      "loss": 1.3382,
+      "step": 4644
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010667558922224872,
+      "loss": 1.2625,
+      "step": 4645
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010664105174919256,
+      "loss": 1.5835,
+      "step": 4646
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010660651348043953,
+      "loss": 1.3743,
+      "step": 4647
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010657197442012787,
+      "loss": 1.1332,
+      "step": 4648
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010653743457239589,
+      "loss": 1.2276,
+      "step": 4649
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010650289394138193,
+      "loss": 1.1651,
+      "step": 4650
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010646835253122449,
+      "loss": 1.2372,
+      "step": 4651
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010643381034606217,
+      "loss": 1.3647,
+      "step": 4652
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010639926739003361,
+      "loss": 1.4755,
+      "step": 4653
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010636472366727756,
+      "loss": 1.6545,
+      "step": 4654
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010633017918193292,
+      "loss": 1.2502,
+      "step": 4655
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010629563393813862,
+      "loss": 1.21,
+      "step": 4656
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010626108794003364,
+      "loss": 1.1444,
+      "step": 4657
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010622654119175715,
+      "loss": 1.1936,
+      "step": 4658
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010619199369744837,
+      "loss": 1.338,
+      "step": 4659
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010615744546124658,
+      "loss": 1.2606,
+      "step": 4660
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010612289648729119,
+      "loss": 1.2441,
+      "step": 4661
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010608834677972166,
+      "loss": 1.4851,
+      "step": 4662
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010605379634267763,
+      "loss": 0.989,
+      "step": 4663
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010601924518029865,
+      "loss": 1.3307,
+      "step": 4664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010598469329672451,
+      "loss": 1.5587,
+      "step": 4665
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010595014069609507,
+      "loss": 1.582,
+      "step": 4666
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010591558738255022,
+      "loss": 1.307,
+      "step": 4667
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010588103336022994,
+      "loss": 1.2807,
+      "step": 4668
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010584647863327437,
+      "loss": 1.6248,
+      "step": 4669
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010581192320582364,
+      "loss": 1.2715,
+      "step": 4670
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000105777367082018,
+      "loss": 1.2706,
+      "step": 4671
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010574281026599779,
+      "loss": 1.5201,
+      "step": 4672
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010570825276190349,
+      "loss": 1.3901,
+      "step": 4673
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010567369457387552,
+      "loss": 1.5384,
+      "step": 4674
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001056391357060545,
+      "loss": 0.9997,
+      "step": 4675
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001056045761625811,
+      "loss": 1.431,
+      "step": 4676
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010557001594759606,
+      "loss": 1.2051,
+      "step": 4677
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010553545506524021,
+      "loss": 1.3415,
+      "step": 4678
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010550089351965448,
+      "loss": 1.3763,
+      "step": 4679
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010546633131497986,
+      "loss": 1.3759,
+      "step": 4680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010543176845535733,
+      "loss": 1.4255,
+      "step": 4681
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010539720494492813,
+      "loss": 1.1635,
+      "step": 4682
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010536264078783349,
+      "loss": 1.4082,
+      "step": 4683
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010532807598821464,
+      "loss": 1.3783,
+      "step": 4684
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010529351055021297,
+      "loss": 1.116,
+      "step": 4685
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010525894447797,
+      "loss": 1.4393,
+      "step": 4686
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001052243777756272,
+      "loss": 1.1717,
+      "step": 4687
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010518981044732616,
+      "loss": 1.4645,
+      "step": 4688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010515524249720863,
+      "loss": 1.3641,
+      "step": 4689
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010512067392941631,
+      "loss": 1.4722,
+      "step": 4690
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010508610474809104,
+      "loss": 1.386,
+      "step": 4691
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010505153495737472,
+      "loss": 1.2916,
+      "step": 4692
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010501696456140937,
+      "loss": 1.2196,
+      "step": 4693
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010498239356433694,
+      "loss": 1.2434,
+      "step": 4694
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010494782197029964,
+      "loss": 1.3998,
+      "step": 4695
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010491324978343966,
+      "loss": 1.5264,
+      "step": 4696
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001048786770078992,
+      "loss": 1.0107,
+      "step": 4697
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010484410364782062,
+      "loss": 1.326,
+      "step": 4698
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010480952970734634,
+      "loss": 1.4165,
+      "step": 4699
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010477495519061884,
+      "loss": 1.2249,
+      "step": 4700
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010474038010178065,
+      "loss": 1.339,
+      "step": 4701
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010470580444497433,
+      "loss": 1.3946,
+      "step": 4702
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010467122822434265,
+      "loss": 1.037,
+      "step": 4703
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010463665144402827,
+      "loss": 1.2054,
+      "step": 4704
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010460207410817406,
+      "loss": 1.2174,
+      "step": 4705
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010456749622092288,
+      "loss": 1.2143,
+      "step": 4706
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001045329177864177,
+      "loss": 1.0039,
+      "step": 4707
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010449833880880146,
+      "loss": 0.9327,
+      "step": 4708
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001044637592922173,
+      "loss": 1.5986,
+      "step": 4709
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010442917924080838,
+      "loss": 1.2711,
+      "step": 4710
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010439459865871783,
+      "loss": 1.353,
+      "step": 4711
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010436001755008893,
+      "loss": 1.2826,
+      "step": 4712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001043254359190651,
+      "loss": 1.2221,
+      "step": 4713
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010429085376978964,
+      "loss": 1.2364,
+      "step": 4714
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010425627110640605,
+      "loss": 1.295,
+      "step": 4715
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010422168793305784,
+      "loss": 1.4505,
+      "step": 4716
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010418710425388857,
+      "loss": 1.342,
+      "step": 4717
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001041525200730419,
+      "loss": 1.3281,
+      "step": 4718
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010411793539466157,
+      "loss": 1.429,
+      "step": 4719
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010408335022289124,
+      "loss": 1.3794,
+      "step": 4720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001040487645618748,
+      "loss": 1.1548,
+      "step": 4721
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001040141784157561,
+      "loss": 1.3489,
+      "step": 4722
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010397959178867911,
+      "loss": 1.1064,
+      "step": 4723
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010394500468478776,
+      "loss": 1.3595,
+      "step": 4724
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010391041710822616,
+      "loss": 1.2939,
+      "step": 4725
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001038758290631384,
+      "loss": 1.3251,
+      "step": 4726
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010384124055366862,
+      "loss": 1.4638,
+      "step": 4727
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010380665158396105,
+      "loss": 1.0975,
+      "step": 4728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010377206215815998,
+      "loss": 0.9529,
+      "step": 4729
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010373747228040973,
+      "loss": 1.3627,
+      "step": 4730
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010370288195485467,
+      "loss": 1.0765,
+      "step": 4731
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010366829118563924,
+      "loss": 1.4052,
+      "step": 4732
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010363369997690795,
+      "loss": 1.193,
+      "step": 4733
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010359910833280529,
+      "loss": 1.5078,
+      "step": 4734
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001035645162574759,
+      "loss": 1.0978,
+      "step": 4735
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001035299237550644,
+      "loss": 1.3048,
+      "step": 4736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010349533082971555,
+      "loss": 1.5262,
+      "step": 4737
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010346073748557401,
+      "loss": 1.3288,
+      "step": 4738
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001034261437267846,
+      "loss": 1.1697,
+      "step": 4739
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010339154955749222,
+      "loss": 1.4018,
+      "step": 4740
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001033569549818417,
+      "loss": 1.1051,
+      "step": 4741
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010332236000397801,
+      "loss": 0.9982,
+      "step": 4742
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010328776462804618,
+      "loss": 1.1103,
+      "step": 4743
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010325316885819118,
+      "loss": 1.1792,
+      "step": 4744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010321857269855817,
+      "loss": 1.2829,
+      "step": 4745
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010318397615329221,
+      "loss": 1.2685,
+      "step": 4746
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010314937922653856,
+      "loss": 1.3008,
+      "step": 4747
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010311478192244235,
+      "loss": 1.4696,
+      "step": 4748
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010308018424514899,
+      "loss": 1.1543,
+      "step": 4749
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010304558619880365,
+      "loss": 1.2103,
+      "step": 4750
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010301098778755178,
+      "loss": 1.4886,
+      "step": 4751
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010297638901553875,
+      "loss": 1.146,
+      "step": 4752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010294178988691006,
+      "loss": 1.2697,
+      "step": 4753
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010290719040581112,
+      "loss": 1.5921,
+      "step": 4754
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010287259057638749,
+      "loss": 0.9281,
+      "step": 4755
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010283799040278481,
+      "loss": 1.2283,
+      "step": 4756
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010280338988914862,
+      "loss": 1.1706,
+      "step": 4757
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010276878903962459,
+      "loss": 1.1578,
+      "step": 4758
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010273418785835842,
+      "loss": 1.1451,
+      "step": 4759
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010269958634949588,
+      "loss": 1.4513,
+      "step": 4760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001026649845171827,
+      "loss": 1.3747,
+      "step": 4761
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001026303823655647,
+      "loss": 1.5787,
+      "step": 4762
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010259577989878777,
+      "loss": 1.1321,
+      "step": 4763
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010256117712099777,
+      "loss": 1.35,
+      "step": 4764
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010252657403634062,
+      "loss": 1.3893,
+      "step": 4765
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001024919706489623,
+      "loss": 1.361,
+      "step": 4766
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010245736696300885,
+      "loss": 1.0831,
+      "step": 4767
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001024227629826262,
+      "loss": 1.0705,
+      "step": 4768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010238815871196052,
+      "loss": 1.2567,
+      "step": 4769
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001023535541551579,
+      "loss": 1.442,
+      "step": 4770
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010231894931636446,
+      "loss": 1.2793,
+      "step": 4771
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010228434419972636,
+      "loss": 1.2321,
+      "step": 4772
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010224973880938986,
+      "loss": 1.2194,
+      "step": 4773
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010221513314950115,
+      "loss": 1.1577,
+      "step": 4774
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001021805272242065,
+      "loss": 0.9209,
+      "step": 4775
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010214592103765225,
+      "loss": 1.2895,
+      "step": 4776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010211131459398476,
+      "loss": 1.2436,
+      "step": 4777
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001020767078973503,
+      "loss": 1.359,
+      "step": 4778
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010204210095189534,
+      "loss": 1.3257,
+      "step": 4779
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010200749376176635,
+      "loss": 1.4041,
+      "step": 4780
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010197288633110968,
+      "loss": 1.2117,
+      "step": 4781
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010193827866407186,
+      "loss": 1.3295,
+      "step": 4782
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010190367076479943,
+      "loss": 1.1309,
+      "step": 4783
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010186906263743889,
+      "loss": 1.2458,
+      "step": 4784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010183445428613685,
+      "loss": 1.3927,
+      "step": 4785
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010179984571503989,
+      "loss": 1.4914,
+      "step": 4786
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010176523692829461,
+      "loss": 1.3687,
+      "step": 4787
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010173062793004768,
+      "loss": 1.4016,
+      "step": 4788
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010169601872444576,
+      "loss": 1.2625,
+      "step": 4789
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010166140931563559,
+      "loss": 1.7872,
+      "step": 4790
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010162679970776382,
+      "loss": 1.0216,
+      "step": 4791
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010159218990497723,
+      "loss": 1.1794,
+      "step": 4792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010155757991142262,
+      "loss": 1.0593,
+      "step": 4793
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010152296973124675,
+      "loss": 1.1656,
+      "step": 4794
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010148835936859645,
+      "loss": 1.1704,
+      "step": 4795
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010145374882761852,
+      "loss": 1.2487,
+      "step": 4796
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010141913811245986,
+      "loss": 1.2975,
+      "step": 4797
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010138452722726734,
+      "loss": 0.7326,
+      "step": 4798
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010134991617618783,
+      "loss": 1.0263,
+      "step": 4799
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001013153049633683,
+      "loss": 1.3779,
+      "step": 4800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010128069359295565,
+      "loss": 0.9717,
+      "step": 4801
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010124608206909682,
+      "loss": 1.1132,
+      "step": 4802
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010121147039593888,
+      "loss": 1.4286,
+      "step": 4803
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010117685857762869,
+      "loss": 1.3693,
+      "step": 4804
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010114224661831335,
+      "loss": 1.6693,
+      "step": 4805
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010110763452213987,
+      "loss": 1.3885,
+      "step": 4806
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010107302229325531,
+      "loss": 1.1808,
+      "step": 4807
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010103840993580667,
+      "loss": 1.4741,
+      "step": 4808
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010100379745394107,
+      "loss": 1.3053,
+      "step": 4809
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010096918485180562,
+      "loss": 1.3248,
+      "step": 4810
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001009345721335474,
+      "loss": 1.4718,
+      "step": 4811
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010089995930331351,
+      "loss": 1.2587,
+      "step": 4812
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010086534636525109,
+      "loss": 1.3227,
+      "step": 4813
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010083073332350734,
+      "loss": 1.5258,
+      "step": 4814
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010079612018222935,
+      "loss": 1.2993,
+      "step": 4815
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001007615069455643,
+      "loss": 1.396,
+      "step": 4816
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010072689361765941,
+      "loss": 1.5889,
+      "step": 4817
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010069228020266183,
+      "loss": 1.5484,
+      "step": 4818
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010065766670471879,
+      "loss": 1.5528,
+      "step": 4819
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010062305312797748,
+      "loss": 1.1486,
+      "step": 4820
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010058843947658513,
+      "loss": 1.4993,
+      "step": 4821
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010055382575468896,
+      "loss": 1.4698,
+      "step": 4822
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010051921196643628,
+      "loss": 1.5229,
+      "step": 4823
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010048459811597424,
+      "loss": 1.2181,
+      "step": 4824
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010044998420745013,
+      "loss": 1.2528,
+      "step": 4825
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010041537024501122,
+      "loss": 1.2953,
+      "step": 4826
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010038075623280478,
+      "loss": 1.3246,
+      "step": 4827
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010034614217497807,
+      "loss": 1.2992,
+      "step": 4828
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010031152807567837,
+      "loss": 1.146,
+      "step": 4829
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010027691393905299,
+      "loss": 1.0212,
+      "step": 4830
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001002422997692492,
+      "loss": 1.1447,
+      "step": 4831
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010020768557041424,
+      "loss": 1.3978,
+      "step": 4832
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010017307134669554,
+      "loss": 1.0648,
+      "step": 4833
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010013845710224026,
+      "loss": 1.2168,
+      "step": 4834
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010010384284119576,
+      "loss": 1.3591,
+      "step": 4835
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010006922856770935,
+      "loss": 1.3922,
+      "step": 4836
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010003461428592836,
+      "loss": 1.3623,
+      "step": 4837
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001,
+      "loss": 1.1264,
+      "step": 4838
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.996538571407168e-05,
+      "loss": 1.4635,
+      "step": 4839
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.993077143229067e-05,
+      "loss": 1.1273,
+      "step": 4840
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.989615715880423e-05,
+      "loss": 1.1797,
+      "step": 4841
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.986154289775976e-05,
+      "loss": 1.1658,
+      "step": 4842
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.98269286533045e-05,
+      "loss": 1.3133,
+      "step": 4843
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.979231442958577e-05,
+      "loss": 1.1847,
+      "step": 4844
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.975770023075081e-05,
+      "loss": 1.4184,
+      "step": 4845
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.972308606094702e-05,
+      "loss": 1.0866,
+      "step": 4846
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.968847192432163e-05,
+      "loss": 1.5359,
+      "step": 4847
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.965385782502196e-05,
+      "loss": 1.403,
+      "step": 4848
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.961924376719521e-05,
+      "loss": 0.9373,
+      "step": 4849
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.958462975498881e-05,
+      "loss": 1.0208,
+      "step": 4850
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.955001579254988e-05,
+      "loss": 1.6291,
+      "step": 4851
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.951540188402579e-05,
+      "loss": 1.3831,
+      "step": 4852
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.948078803356374e-05,
+      "loss": 1.5353,
+      "step": 4853
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.944617424531105e-05,
+      "loss": 1.2807,
+      "step": 4854
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.941156052341487e-05,
+      "loss": 1.4444,
+      "step": 4855
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.937694687202255e-05,
+      "loss": 1.3574,
+      "step": 4856
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.934233329528124e-05,
+      "loss": 1.483,
+      "step": 4857
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.93077197973382e-05,
+      "loss": 1.2173,
+      "step": 4858
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.92731063823406e-05,
+      "loss": 1.1366,
+      "step": 4859
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.923849305443573e-05,
+      "loss": 1.149,
+      "step": 4860
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.920387981777067e-05,
+      "loss": 1.4594,
+      "step": 4861
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.91692666764927e-05,
+      "loss": 1.5395,
+      "step": 4862
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.913465363474892e-05,
+      "loss": 1.3896,
+      "step": 4863
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.910004069668655e-05,
+      "loss": 1.3974,
+      "step": 4864
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.906542786645264e-05,
+      "loss": 1.1701,
+      "step": 4865
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.90308151481944e-05,
+      "loss": 1.6608,
+      "step": 4866
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.899620254605894e-05,
+      "loss": 0.9824,
+      "step": 4867
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.896159006419338e-05,
+      "loss": 1.1813,
+      "step": 4868
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.892697770674473e-05,
+      "loss": 1.2482,
+      "step": 4869
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.889236547786016e-05,
+      "loss": 1.3859,
+      "step": 4870
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.885775338168667e-05,
+      "loss": 1.1742,
+      "step": 4871
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.882314142237134e-05,
+      "loss": 1.4071,
+      "step": 4872
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.878852960406116e-05,
+      "loss": 1.2504,
+      "step": 4873
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.875391793090321e-05,
+      "loss": 1.3275,
+      "step": 4874
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.871930640704438e-05,
+      "loss": 1.3345,
+      "step": 4875
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.868469503663172e-05,
+      "loss": 1.4583,
+      "step": 4876
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.865008382381219e-05,
+      "loss": 1.3097,
+      "step": 4877
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.861547277273267e-05,
+      "loss": 1.039,
+      "step": 4878
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.858086188754016e-05,
+      "loss": 1.0784,
+      "step": 4879
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.854625117238149e-05,
+      "loss": 1.184,
+      "step": 4880
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.851164063140359e-05,
+      "loss": 1.1175,
+      "step": 4881
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.847703026875324e-05,
+      "loss": 1.2084,
+      "step": 4882
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.844242008857739e-05,
+      "loss": 1.2991,
+      "step": 4883
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.840781009502276e-05,
+      "loss": 1.0941,
+      "step": 4884
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.83732002922362e-05,
+      "loss": 1.351,
+      "step": 4885
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.833859068436442e-05,
+      "loss": 1.275,
+      "step": 4886
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.830398127555426e-05,
+      "loss": 1.2996,
+      "step": 4887
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.826937206995233e-05,
+      "loss": 1.2086,
+      "step": 4888
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.82347630717054e-05,
+      "loss": 1.6298,
+      "step": 4889
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.820015428496013e-05,
+      "loss": 1.1285,
+      "step": 4890
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.816554571386317e-05,
+      "loss": 1.3285,
+      "step": 4891
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.81309373625611e-05,
+      "loss": 1.471,
+      "step": 4892
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.80963292352006e-05,
+      "loss": 0.7381,
+      "step": 4893
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.806172133592816e-05,
+      "loss": 1.141,
+      "step": 4894
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.802711366889037e-05,
+      "loss": 1.3078,
+      "step": 4895
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.799250623823369e-05,
+      "loss": 1.2516,
+      "step": 4896
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.795789904810466e-05,
+      "loss": 1.4962,
+      "step": 4897
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.792329210264972e-05,
+      "loss": 1.2858,
+      "step": 4898
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.78886854060153e-05,
+      "loss": 1.4034,
+      "step": 4899
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.785407896234777e-05,
+      "loss": 1.2006,
+      "step": 4900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.781947277579354e-05,
+      "loss": 1.2553,
+      "step": 4901
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.778486685049888e-05,
+      "loss": 1.477,
+      "step": 4902
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.775026119061019e-05,
+      "loss": 1.2471,
+      "step": 4903
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.771565580027366e-05,
+      "loss": 1.2031,
+      "step": 4904
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.76810506836356e-05,
+      "loss": 1.3454,
+      "step": 4905
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.76464458448421e-05,
+      "loss": 1.2904,
+      "step": 4906
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.761184128803949e-05,
+      "loss": 1.2823,
+      "step": 4907
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.75772370173738e-05,
+      "loss": 1.3606,
+      "step": 4908
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.75426330369912e-05,
+      "loss": 1.2585,
+      "step": 4909
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.750802935103772e-05,
+      "loss": 1.3087,
+      "step": 4910
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.747342596365938e-05,
+      "loss": 1.5485,
+      "step": 4911
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.743882287900225e-05,
+      "loss": 1.4052,
+      "step": 4912
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.740422010121224e-05,
+      "loss": 1.183,
+      "step": 4913
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.736961763443532e-05,
+      "loss": 1.0775,
+      "step": 4914
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.73350154828173e-05,
+      "loss": 1.1532,
+      "step": 4915
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.730041365050414e-05,
+      "loss": 1.2692,
+      "step": 4916
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.726581214164159e-05,
+      "loss": 1.4164,
+      "step": 4917
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.723121096037543e-05,
+      "loss": 1.4155,
+      "step": 4918
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.719661011085139e-05,
+      "loss": 1.2689,
+      "step": 4919
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.716200959721521e-05,
+      "loss": 1.4484,
+      "step": 4920
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.712740942361249e-05,
+      "loss": 1.4423,
+      "step": 4921
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.70928095941889e-05,
+      "loss": 1.557,
+      "step": 4922
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.705821011308998e-05,
+      "loss": 1.3379,
+      "step": 4923
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.702361098446126e-05,
+      "loss": 1.2219,
+      "step": 4924
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.698901221244823e-05,
+      "loss": 1.5438,
+      "step": 4925
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.695441380119637e-05,
+      "loss": 1.5947,
+      "step": 4926
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.691981575485105e-05,
+      "loss": 1.6005,
+      "step": 4927
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.688521807755766e-05,
+      "loss": 1.3938,
+      "step": 4928
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.685062077346147e-05,
+      "loss": 1.4459,
+      "step": 4929
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.68160238467078e-05,
+      "loss": 1.3288,
+      "step": 4930
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.678142730144186e-05,
+      "loss": 1.1118,
+      "step": 4931
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.674683114180884e-05,
+      "loss": 1.2775,
+      "step": 4932
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.671223537195383e-05,
+      "loss": 1.3517,
+      "step": 4933
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.667763999602202e-05,
+      "loss": 1.5223,
+      "step": 4934
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.664304501815831e-05,
+      "loss": 1.6801,
+      "step": 4935
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.660845044250782e-05,
+      "loss": 1.6704,
+      "step": 4936
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.657385627321542e-05,
+      "loss": 1.246,
+      "step": 4937
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.653926251442604e-05,
+      "loss": 1.2285,
+      "step": 4938
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.650466917028447e-05,
+      "loss": 0.5332,
+      "step": 4939
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.647007624493562e-05,
+      "loss": 1.1914,
+      "step": 4940
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.643548374252413e-05,
+      "loss": 1.305,
+      "step": 4941
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.640089166719475e-05,
+      "loss": 1.138,
+      "step": 4942
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.636630002309208e-05,
+      "loss": 1.4861,
+      "step": 4943
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.633170881436079e-05,
+      "loss": 1.4879,
+      "step": 4944
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.629711804514536e-05,
+      "loss": 1.0451,
+      "step": 4945
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.626252771959031e-05,
+      "loss": 1.0983,
+      "step": 4946
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.622793784184005e-05,
+      "loss": 1.5124,
+      "step": 4947
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.619334841603895e-05,
+      "loss": 1.4916,
+      "step": 4948
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.615875944633141e-05,
+      "loss": 1.2912,
+      "step": 4949
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.612417093686162e-05,
+      "loss": 1.1811,
+      "step": 4950
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.608958289177386e-05,
+      "loss": 1.3297,
+      "step": 4951
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.605499531521223e-05,
+      "loss": 1.2086,
+      "step": 4952
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.602040821132093e-05,
+      "loss": 1.3065,
+      "step": 4953
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.598582158424391e-05,
+      "loss": 1.3957,
+      "step": 4954
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.595123543812523e-05,
+      "loss": 1.1653,
+      "step": 4955
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.591664977710875e-05,
+      "loss": 1.4288,
+      "step": 4956
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.588206460533847e-05,
+      "loss": 1.3365,
+      "step": 4957
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.584747992695808e-05,
+      "loss": 1.209,
+      "step": 4958
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.581289574611144e-05,
+      "loss": 1.1795,
+      "step": 4959
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.577831206694218e-05,
+      "loss": 1.3344,
+      "step": 4960
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.574372889359398e-05,
+      "loss": 1.4933,
+      "step": 4961
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.570914623021036e-05,
+      "loss": 1.1791,
+      "step": 4962
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.567456408093492e-05,
+      "loss": 1.2287,
+      "step": 4963
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.563998244991105e-05,
+      "loss": 1.1969,
+      "step": 4964
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.560540134128222e-05,
+      "loss": 1.1578,
+      "step": 4965
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.557082075919164e-05,
+      "loss": 1.0698,
+      "step": 4966
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.553624070778271e-05,
+      "loss": 0.8423,
+      "step": 4967
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.550166119119856e-05,
+      "loss": 1.4298,
+      "step": 4968
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.546708221358236e-05,
+      "loss": 1.1554,
+      "step": 4969
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.543250377907715e-05,
+      "loss": 1.0173,
+      "step": 4970
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.539792589182597e-05,
+      "loss": 1.0576,
+      "step": 4971
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.536334855597174e-05,
+      "loss": 1.0149,
+      "step": 4972
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.53287717756574e-05,
+      "loss": 1.4422,
+      "step": 4973
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.529419555502569e-05,
+      "loss": 1.177,
+      "step": 4974
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.525961989821942e-05,
+      "loss": 1.648,
+      "step": 4975
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.522504480938116e-05,
+      "loss": 1.2899,
+      "step": 4976
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.519047029265367e-05,
+      "loss": 1.5737,
+      "step": 4977
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.51558963521794e-05,
+      "loss": 1.5401,
+      "step": 4978
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.512132299210084e-05,
+      "loss": 1.4744,
+      "step": 4979
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.508675021656036e-05,
+      "loss": 1.3869,
+      "step": 4980
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.50521780297004e-05,
+      "loss": 1.1667,
+      "step": 4981
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.501760643566307e-05,
+      "loss": 1.4073,
+      "step": 4982
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.498303543859066e-05,
+      "loss": 1.1401,
+      "step": 4983
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.49484650426253e-05,
+      "loss": 1.4092,
+      "step": 4984
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.491389525190896e-05,
+      "loss": 1.083,
+      "step": 4985
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.487932607058371e-05,
+      "loss": 1.0905,
+      "step": 4986
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.484475750279138e-05,
+      "loss": 1.2716,
+      "step": 4987
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.481018955267386e-05,
+      "loss": 1.2319,
+      "step": 4988
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.47756222243728e-05,
+      "loss": 1.3982,
+      "step": 4989
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.474105552203001e-05,
+      "loss": 1.3665,
+      "step": 4990
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.470648944978701e-05,
+      "loss": 1.15,
+      "step": 4991
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.46719240117854e-05,
+      "loss": 1.2094,
+      "step": 4992
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.46373592121665e-05,
+      "loss": 1.2999,
+      "step": 4993
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.460279505507188e-05,
+      "loss": 1.2671,
+      "step": 4994
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.456823154464266e-05,
+      "loss": 1.2977,
+      "step": 4995
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.453366868502018e-05,
+      "loss": 1.3414,
+      "step": 4996
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.449910648034553e-05,
+      "loss": 1.5032,
+      "step": 4997
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.44645449347598e-05,
+      "loss": 1.5557,
+      "step": 4998
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.442998405240395e-05,
+      "loss": 1.3107,
+      "step": 4999
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.439542383741893e-05,
+      "loss": 1.6359,
+      "step": 5000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.436086429394552e-05,
+      "loss": 0.9483,
+      "step": 5001
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.432630542612453e-05,
+      "loss": 1.6296,
+      "step": 5002
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.429174723809652e-05,
+      "loss": 1.2302,
+      "step": 5003
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.425718973400222e-05,
+      "loss": 1.2819,
+      "step": 5004
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.422263291798203e-05,
+      "loss": 1.2842,
+      "step": 5005
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.41880767941764e-05,
+      "loss": 1.3535,
+      "step": 5006
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.415352136672566e-05,
+      "loss": 1.3575,
+      "step": 5007
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.41189666397701e-05,
+      "loss": 1.3405,
+      "step": 5008
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.40844126174498e-05,
+      "loss": 1.1782,
+      "step": 5009
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.404985930390496e-05,
+      "loss": 1.511,
+      "step": 5010
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.40153067032755e-05,
+      "loss": 1.2666,
+      "step": 5011
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.39807548197014e-05,
+      "loss": 1.3498,
+      "step": 5012
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.394620365732241e-05,
+      "loss": 1.1758,
+      "step": 5013
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.391165322027835e-05,
+      "loss": 1.3149,
+      "step": 5014
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.387710351270883e-05,
+      "loss": 1.2024,
+      "step": 5015
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.384255453875346e-05,
+      "loss": 1.6865,
+      "step": 5016
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.380800630255164e-05,
+      "loss": 1.4341,
+      "step": 5017
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.37734588082429e-05,
+      "loss": 1.2785,
+      "step": 5018
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.373891205996639e-05,
+      "loss": 1.0934,
+      "step": 5019
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.370436606186142e-05,
+      "loss": 1.4041,
+      "step": 5020
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.36698208180671e-05,
+      "loss": 1.6417,
+      "step": 5021
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.363527633272242e-05,
+      "loss": 1.1118,
+      "step": 5022
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.360073260996641e-05,
+      "loss": 1.2768,
+      "step": 5023
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.356618965393784e-05,
+      "loss": 1.5173,
+      "step": 5024
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.353164746877553e-05,
+      "loss": 1.299,
+      "step": 5025
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.349710605861807e-05,
+      "loss": 1.4306,
+      "step": 5026
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.346256542760413e-05,
+      "loss": 1.0263,
+      "step": 5027
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.342802557987212e-05,
+      "loss": 1.1011,
+      "step": 5028
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.339348651956048e-05,
+      "loss": 1.5228,
+      "step": 5029
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.335894825080746e-05,
+      "loss": 1.5286,
+      "step": 5030
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.33244107777513e-05,
+      "loss": 1.295,
+      "step": 5031
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.328987410453002e-05,
+      "loss": 1.0525,
+      "step": 5032
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.325533823528174e-05,
+      "loss": 1.1574,
+      "step": 5033
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.32208031741443e-05,
+      "loss": 1.0116,
+      "step": 5034
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.318626892525554e-05,
+      "loss": 1.1709,
+      "step": 5035
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.315173549275313e-05,
+      "loss": 0.7826,
+      "step": 5036
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.311720288077478e-05,
+      "loss": 1.4218,
+      "step": 5037
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.308267109345793e-05,
+      "loss": 1.3226,
+      "step": 5038
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.304814013494005e-05,
+      "loss": 1.2743,
+      "step": 5039
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.301361000935841e-05,
+      "loss": 1.1424,
+      "step": 5040
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.297908072085032e-05,
+      "loss": 1.2954,
+      "step": 5041
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.29445522735528e-05,
+      "loss": 1.5263,
+      "step": 5042
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.291002467160295e-05,
+      "loss": 1.2829,
+      "step": 5043
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.287549791913765e-05,
+      "loss": 1.4464,
+      "step": 5044
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.284097202029375e-05,
+      "loss": 1.1671,
+      "step": 5045
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.280644697920791e-05,
+      "loss": 1.2668,
+      "step": 5046
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.27719228000168e-05,
+      "loss": 1.2235,
+      "step": 5047
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.273739948685691e-05,
+      "loss": 1.1693,
+      "step": 5048
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.270287704386467e-05,
+      "loss": 1.6236,
+      "step": 5049
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.266835547517631e-05,
+      "loss": 1.3179,
+      "step": 5050
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.263383478492813e-05,
+      "loss": 1.3689,
+      "step": 5051
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.259931497725614e-05,
+      "loss": 1.6001,
+      "step": 5052
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.256479605629639e-05,
+      "loss": 1.3733,
+      "step": 5053
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.253027802618469e-05,
+      "loss": 1.2204,
+      "step": 5054
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.249576089105684e-05,
+      "loss": 1.5029,
+      "step": 5055
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.246124465504856e-05,
+      "loss": 1.6464,
+      "step": 5056
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.242672932229535e-05,
+      "loss": 1.4171,
+      "step": 5057
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.239221489693268e-05,
+      "loss": 1.5658,
+      "step": 5058
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.235770138309586e-05,
+      "loss": 1.1262,
+      "step": 5059
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.23231887849202e-05,
+      "loss": 1.2506,
+      "step": 5060
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.228867710654074e-05,
+      "loss": 1.0334,
+      "step": 5061
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.225416635209256e-05,
+      "loss": 1.6442,
+      "step": 5062
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.221965652571048e-05,
+      "loss": 1.3052,
+      "step": 5063
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.218514763152938e-05,
+      "loss": 0.9823,
+      "step": 5064
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.215063967368389e-05,
+      "loss": 1.0563,
+      "step": 5065
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.211613265630861e-05,
+      "loss": 1.3876,
+      "step": 5066
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.208162658353795e-05,
+      "loss": 1.2308,
+      "step": 5067
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.204712145950627e-05,
+      "loss": 1.546,
+      "step": 5068
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.201261728834777e-05,
+      "loss": 1.2964,
+      "step": 5069
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.197811407419663e-05,
+      "loss": 1.1073,
+      "step": 5070
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.19436118211868e-05,
+      "loss": 0.9877,
+      "step": 5071
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.190911053345219e-05,
+      "loss": 1.5507,
+      "step": 5072
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.18746102151265e-05,
+      "loss": 1.2598,
+      "step": 5073
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.184011087034348e-05,
+      "loss": 1.1412,
+      "step": 5074
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.18056125032366e-05,
+      "loss": 1.2098,
+      "step": 5075
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.177111511793932e-05,
+      "loss": 1.5989,
+      "step": 5076
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.173661871858487e-05,
+      "loss": 1.0569,
+      "step": 5077
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.170212330930654e-05,
+      "loss": 1.3787,
+      "step": 5078
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.166762889423728e-05,
+      "loss": 1.3391,
+      "step": 5079
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.163313547751012e-05,
+      "loss": 1.1267,
+      "step": 5080
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.159864306325782e-05,
+      "loss": 1.3914,
+      "step": 5081
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.156415165561315e-05,
+      "loss": 1.2963,
+      "step": 5082
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.152966125870861e-05,
+      "loss": 1.0828,
+      "step": 5083
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.149517187667675e-05,
+      "loss": 0.9465,
+      "step": 5084
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.146068351364986e-05,
+      "loss": 1.2305,
+      "step": 5085
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.142619617376021e-05,
+      "loss": 1.0864,
+      "step": 5086
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.139170986113978e-05,
+      "loss": 1.2295,
+      "step": 5087
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.135722457992071e-05,
+      "loss": 0.9754,
+      "step": 5088
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.13227403342347e-05,
+      "loss": 0.9092,
+      "step": 5089
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.128825712821358e-05,
+      "loss": 1.5597,
+      "step": 5090
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.125377496598888e-05,
+      "loss": 1.0901,
+      "step": 5091
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.121929385169209e-05,
+      "loss": 0.9685,
+      "step": 5092
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.118481378945462e-05,
+      "loss": 1.1748,
+      "step": 5093
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.115033478340763e-05,
+      "loss": 1.4694,
+      "step": 5094
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.111585683768225e-05,
+      "loss": 1.4026,
+      "step": 5095
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.108137995640941e-05,
+      "loss": 1.2828,
+      "step": 5096
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.104690414372001e-05,
+      "loss": 1.2994,
+      "step": 5097
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.101242940374474e-05,
+      "loss": 0.7576,
+      "step": 5098
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.09779557406142e-05,
+      "loss": 1.1465,
+      "step": 5099
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.094348315845877e-05,
+      "loss": 1.1891,
+      "step": 5100
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.090901166140893e-05,
+      "loss": 1.3101,
+      "step": 5101
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.087454125359473e-05,
+      "loss": 1.1316,
+      "step": 5102
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.084007193914632e-05,
+      "loss": 0.9022,
+      "step": 5103
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.080560372219361e-05,
+      "loss": 1.283,
+      "step": 5104
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.077113660686642e-05,
+      "loss": 1.2,
+      "step": 5105
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.073667059729439e-05,
+      "loss": 1.3313,
+      "step": 5106
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.070220569760713e-05,
+      "loss": 1.3018,
+      "step": 5107
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.066774191193395e-05,
+      "loss": 1.3642,
+      "step": 5108
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.063327924440422e-05,
+      "loss": 1.0692,
+      "step": 5109
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.0598817699147e-05,
+      "loss": 1.1946,
+      "step": 5110
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.056435728029137e-05,
+      "loss": 1.204,
+      "step": 5111
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.052989799196613e-05,
+      "loss": 1.37,
+      "step": 5112
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.04954398383001e-05,
+      "loss": 1.5322,
+      "step": 5113
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.046098282342179e-05,
+      "loss": 1.2791,
+      "step": 5114
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.042652695145974e-05,
+      "loss": 1.3215,
+      "step": 5115
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.039207222654218e-05,
+      "loss": 1.5261,
+      "step": 5116
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.035761865279741e-05,
+      "loss": 1.3852,
+      "step": 5117
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.03231662343534e-05,
+      "loss": 1.1737,
+      "step": 5118
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.028871497533812e-05,
+      "loss": 1.1579,
+      "step": 5119
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.025426487987927e-05,
+      "loss": 1.486,
+      "step": 5120
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.021981595210457e-05,
+      "loss": 1.7507,
+      "step": 5121
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.018536819614145e-05,
+      "loss": 1.2168,
+      "step": 5122
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.015092161611731e-05,
+      "loss": 1.1244,
+      "step": 5123
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.011647621615931e-05,
+      "loss": 1.3217,
+      "step": 5124
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.00820320003946e-05,
+      "loss": 1.2566,
+      "step": 5125
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.004758897295e-05,
+      "loss": 1.3332,
+      "step": 5126
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.00131471379524e-05,
+      "loss": 1.2235,
+      "step": 5127
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.99787064995284e-05,
+      "loss": 1.2326,
+      "step": 5128
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.994426706180448e-05,
+      "loss": 1.4176,
+      "step": 5129
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.990982882890705e-05,
+      "loss": 1.4251,
+      "step": 5130
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.987539180496228e-05,
+      "loss": 1.1647,
+      "step": 5131
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.984095599409627e-05,
+      "loss": 1.0247,
+      "step": 5132
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.98065214004349e-05,
+      "loss": 1.3243,
+      "step": 5133
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.977208802810399e-05,
+      "loss": 1.2832,
+      "step": 5134
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.973765588122914e-05,
+      "loss": 0.9345,
+      "step": 5135
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.970322496393587e-05,
+      "loss": 1.2923,
+      "step": 5136
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.966879528034945e-05,
+      "loss": 1.401,
+      "step": 5137
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.963436683459517e-05,
+      "loss": 1.1828,
+      "step": 5138
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.959993963079795e-05,
+      "loss": 1.6227,
+      "step": 5139
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.956551367308279e-05,
+      "loss": 1.1891,
+      "step": 5140
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.953108896557436e-05,
+      "loss": 0.9775,
+      "step": 5141
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.94966655123973e-05,
+      "loss": 0.8462,
+      "step": 5142
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.946224331767599e-05,
+      "loss": 1.3398,
+      "step": 5143
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.942782238553479e-05,
+      "loss": 1.2426,
+      "step": 5144
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.939340272009779e-05,
+      "loss": 1.2985,
+      "step": 5145
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.935898432548903e-05,
+      "loss": 1.2951,
+      "step": 5146
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.932456720583226e-05,
+      "loss": 1.0929,
+      "step": 5147
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.929015136525129e-05,
+      "loss": 1.5432,
+      "step": 5148
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.92557368078695e-05,
+      "loss": 1.417,
+      "step": 5149
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.92213235378104e-05,
+      "loss": 1.1421,
+      "step": 5150
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.918691155919713e-05,
+      "loss": 1.5085,
+      "step": 5151
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.91525008761528e-05,
+      "loss": 1.3919,
+      "step": 5152
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.911809149280026e-05,
+      "loss": 1.2834,
+      "step": 5153
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.908368341326237e-05,
+      "loss": 0.9233,
+      "step": 5154
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.904927664166163e-05,
+      "loss": 1.0985,
+      "step": 5155
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.901487118212056e-05,
+      "loss": 1.3084,
+      "step": 5156
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.898046703876135e-05,
+      "loss": 1.4965,
+      "step": 5157
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.894606421570624e-05,
+      "loss": 1.496,
+      "step": 5158
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.891166271707713e-05,
+      "loss": 1.4962,
+      "step": 5159
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.887726254699586e-05,
+      "loss": 1.0831,
+      "step": 5160
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.884286370958407e-05,
+      "loss": 1.2915,
+      "step": 5161
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.880846620896326e-05,
+      "loss": 1.3096,
+      "step": 5162
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.877407004925471e-05,
+      "loss": 1.6443,
+      "step": 5163
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.873967523457968e-05,
+      "loss": 0.9668,
+      "step": 5164
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.870528176905916e-05,
+      "loss": 1.2852,
+      "step": 5165
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.867088965681393e-05,
+      "loss": 1.1507,
+      "step": 5166
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.863649890196475e-05,
+      "loss": 1.3775,
+      "step": 5167
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.860210950863212e-05,
+      "loss": 1.3188,
+      "step": 5168
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.856772148093643e-05,
+      "loss": 1.0937,
+      "step": 5169
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.853333482299779e-05,
+      "loss": 1.3297,
+      "step": 5170
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.849894953893634e-05,
+      "loss": 1.6426,
+      "step": 5171
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.846456563287189e-05,
+      "loss": 1.4546,
+      "step": 5172
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.843018310892416e-05,
+      "loss": 1.2651,
+      "step": 5173
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.839580197121266e-05,
+      "loss": 1.118,
+      "step": 5174
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.836142222385682e-05,
+      "loss": 1.4293,
+      "step": 5175
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.832704387097576e-05,
+      "loss": 1.2514,
+      "step": 5176
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.829266691668861e-05,
+      "loss": 1.263,
+      "step": 5177
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.825829136511419e-05,
+      "loss": 1.1466,
+      "step": 5178
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.822391722037123e-05,
+      "loss": 1.2819,
+      "step": 5179
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.818954448657819e-05,
+      "loss": 1.2838,
+      "step": 5180
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.815517316785354e-05,
+      "loss": 1.2859,
+      "step": 5181
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.81208032683154e-05,
+      "loss": 1.0728,
+      "step": 5182
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.808643479208185e-05,
+      "loss": 1.4116,
+      "step": 5183
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.805206774327067e-05,
+      "loss": 1.3995,
+      "step": 5184
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.801770212599966e-05,
+      "loss": 1.3151,
+      "step": 5185
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.798333794438621e-05,
+      "loss": 1.2866,
+      "step": 5186
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.794897520254774e-05,
+      "loss": 1.2363,
+      "step": 5187
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.791461390460136e-05,
+      "loss": 1.488,
+      "step": 5188
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.788025405466414e-05,
+      "loss": 1.1124,
+      "step": 5189
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.784589565685281e-05,
+      "loss": 1.6841,
+      "step": 5190
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.781153871528411e-05,
+      "loss": 0.86,
+      "step": 5191
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.777718323407444e-05,
+      "loss": 1.1572,
+      "step": 5192
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.774282921734014e-05,
+      "loss": 1.6635,
+      "step": 5193
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.770847666919729e-05,
+      "loss": 1.5467,
+      "step": 5194
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.76741255937619e-05,
+      "loss": 1.2846,
+      "step": 5195
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.763977599514968e-05,
+      "loss": 1.2696,
+      "step": 5196
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.760542787747629e-05,
+      "loss": 1.5227,
+      "step": 5197
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.757108124485707e-05,
+      "loss": 1.2555,
+      "step": 5198
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.753673610140726e-05,
+      "loss": 1.162,
+      "step": 5199
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.750239245124201e-05,
+      "loss": 1.5653,
+      "step": 5200
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.746805029847611e-05,
+      "loss": 1.1794,
+      "step": 5201
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.743370964722431e-05,
+      "loss": 1.1551,
+      "step": 5202
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.739937050160109e-05,
+      "loss": 1.1789,
+      "step": 5203
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.736503286572085e-05,
+      "loss": 1.0609,
+      "step": 5204
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.733069674369767e-05,
+      "loss": 1.5263,
+      "step": 5205
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.729636213964562e-05,
+      "loss": 1.1573,
+      "step": 5206
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.72620290576784e-05,
+      "loss": 1.3372,
+      "step": 5207
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.722769750190973e-05,
+      "loss": 1.4547,
+      "step": 5208
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.719336747645293e-05,
+      "loss": 1.6422,
+      "step": 5209
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.715903898542133e-05,
+      "loss": 1.2009,
+      "step": 5210
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.712471203292797e-05,
+      "loss": 0.9282,
+      "step": 5211
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.709038662308573e-05,
+      "loss": 1.0775,
+      "step": 5212
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.705606276000726e-05,
+      "loss": 1.4072,
+      "step": 5213
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.702174044780517e-05,
+      "loss": 1.1951,
+      "step": 5214
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.698741969059169e-05,
+      "loss": 1.3745,
+      "step": 5215
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.695310049247901e-05,
+      "loss": 0.8802,
+      "step": 5216
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.691878285757901e-05,
+      "loss": 1.2257,
+      "step": 5217
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.688446679000357e-05,
+      "loss": 1.0353,
+      "step": 5218
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.685015229386419e-05,
+      "loss": 1.3735,
+      "step": 5219
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.681583937327226e-05,
+      "loss": 0.8999,
+      "step": 5220
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.678152803233899e-05,
+      "loss": 1.472,
+      "step": 5221
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.67472182751754e-05,
+      "loss": 1.5747,
+      "step": 5222
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.671291010589225e-05,
+      "loss": 1.4728,
+      "step": 5223
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.667860352860026e-05,
+      "loss": 1.2549,
+      "step": 5224
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.664429854740982e-05,
+      "loss": 1.292,
+      "step": 5225
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.660999516643118e-05,
+      "loss": 1.4879,
+      "step": 5226
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.657569338977437e-05,
+      "loss": 1.0526,
+      "step": 5227
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.654139322154931e-05,
+      "loss": 1.406,
+      "step": 5228
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.650709466586564e-05,
+      "loss": 1.4706,
+      "step": 5229
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.647279772683284e-05,
+      "loss": 1.2592,
+      "step": 5230
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.643850240856016e-05,
+      "loss": 1.776,
+      "step": 5231
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.640420871515679e-05,
+      "loss": 1.2952,
+      "step": 5232
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.636991665073149e-05,
+      "loss": 1.3159,
+      "step": 5233
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.633562621939306e-05,
+      "loss": 1.431,
+      "step": 5234
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.630133742524996e-05,
+      "loss": 1.3462,
+      "step": 5235
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.626705027241048e-05,
+      "loss": 1.2467,
+      "step": 5236
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.623276476498281e-05,
+      "loss": 1.2381,
+      "step": 5237
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.619848090707479e-05,
+      "loss": 1.1282,
+      "step": 5238
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.616419870279418e-05,
+      "loss": 1.126,
+      "step": 5239
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.612991815624842e-05,
+      "loss": 1.4206,
+      "step": 5240
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.609563927154496e-05,
+      "loss": 1.5395,
+      "step": 5241
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.606136205279081e-05,
+      "loss": 1.1541,
+      "step": 5242
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.602708650409295e-05,
+      "loss": 1.2478,
+      "step": 5243
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.599281262955804e-05,
+      "loss": 1.1507,
+      "step": 5244
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.595854043329271e-05,
+      "loss": 0.7792,
+      "step": 5245
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.592426991940314e-05,
+      "loss": 1.337,
+      "step": 5246
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.589000109199556e-05,
+      "loss": 1.5053,
+      "step": 5247
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.585573395517583e-05,
+      "loss": 1.4976,
+      "step": 5248
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.582146851304969e-05,
+      "loss": 1.7961,
+      "step": 5249
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.578720476972259e-05,
+      "loss": 1.0869,
+      "step": 5250
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.575294272929994e-05,
+      "loss": 1.0945,
+      "step": 5251
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.571868239588674e-05,
+      "loss": 1.242,
+      "step": 5252
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.568442377358797e-05,
+      "loss": 1.0103,
+      "step": 5253
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.565016686650824e-05,
+      "loss": 1.4739,
+      "step": 5254
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.561591167875213e-05,
+      "loss": 1.1095,
+      "step": 5255
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.558165821442385e-05,
+      "loss": 1.2717,
+      "step": 5256
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.554740647762752e-05,
+      "loss": 1.0373,
+      "step": 5257
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.551315647246695e-05,
+      "loss": 1.2347,
+      "step": 5258
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.547890820304587e-05,
+      "loss": 1.6584,
+      "step": 5259
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.544466167346765e-05,
+      "loss": 1.2332,
+      "step": 5260
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.541041688783564e-05,
+      "loss": 1.1742,
+      "step": 5261
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.537617385025279e-05,
+      "loss": 1.3983,
+      "step": 5262
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.534193256482196e-05,
+      "loss": 1.292,
+      "step": 5263
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.530769303564575e-05,
+      "loss": 1.1197,
+      "step": 5264
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.527345526682658e-05,
+      "loss": 1.3803,
+      "step": 5265
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.523921926246664e-05,
+      "loss": 1.253,
+      "step": 5266
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.520498502666793e-05,
+      "loss": 1.537,
+      "step": 5267
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.517075256353217e-05,
+      "loss": 1.236,
+      "step": 5268
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.513652187716099e-05,
+      "loss": 1.3931,
+      "step": 5269
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.510229297165564e-05,
+      "loss": 1.4244,
+      "step": 5270
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.506806585111733e-05,
+      "loss": 1.0477,
+      "step": 5271
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.503384051964699e-05,
+      "loss": 1.2539,
+      "step": 5272
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.499961698134524e-05,
+      "loss": 1.2267,
+      "step": 5273
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.496539524031266e-05,
+      "loss": 1.3112,
+      "step": 5274
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.493117530064948e-05,
+      "loss": 1.0737,
+      "step": 5275
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.489695716645578e-05,
+      "loss": 1.4066,
+      "step": 5276
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.486274084183132e-05,
+      "loss": 1.2119,
+      "step": 5277
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.482852633087585e-05,
+      "loss": 1.0846,
+      "step": 5278
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.47943136376887e-05,
+      "loss": 1.4399,
+      "step": 5279
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.476010276636909e-05,
+      "loss": 1.2901,
+      "step": 5280
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.472589372101595e-05,
+      "loss": 1.5316,
+      "step": 5281
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.46916865057281e-05,
+      "loss": 1.2663,
+      "step": 5282
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.465748112460397e-05,
+      "loss": 1.2985,
+      "step": 5283
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.4623277581742e-05,
+      "loss": 1.2019,
+      "step": 5284
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.458907588124019e-05,
+      "loss": 1.4494,
+      "step": 5285
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.455487602719647e-05,
+      "loss": 1.4336,
+      "step": 5286
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.452067802370841e-05,
+      "loss": 1.5281,
+      "step": 5287
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.448648187487356e-05,
+      "loss": 1.1976,
+      "step": 5288
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.445228758478903e-05,
+      "loss": 1.2106,
+      "step": 5289
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.441809515755186e-05,
+      "loss": 1.4367,
+      "step": 5290
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.438390459725875e-05,
+      "loss": 1.2558,
+      "step": 5291
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.434971590800633e-05,
+      "loss": 1.3338,
+      "step": 5292
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.43155290938908e-05,
+      "loss": 1.335,
+      "step": 5293
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.428134415900837e-05,
+      "loss": 1.158,
+      "step": 5294
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.42471611074548e-05,
+      "loss": 1.2678,
+      "step": 5295
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.421297994332583e-05,
+      "loss": 1.4017,
+      "step": 5296
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.417880067071675e-05,
+      "loss": 1.5021,
+      "step": 5297
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.414462329372284e-05,
+      "loss": 0.9569,
+      "step": 5298
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.411044781643902e-05,
+      "loss": 1.3381,
+      "step": 5299
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.407627424296005e-05,
+      "loss": 1.3039,
+      "step": 5300
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.404210257738038e-05,
+      "loss": 1.4023,
+      "step": 5301
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.400793282379436e-05,
+      "loss": 1.2985,
+      "step": 5302
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.397376498629594e-05,
+      "loss": 1.3985,
+      "step": 5303
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.393959906897903e-05,
+      "loss": 1.2376,
+      "step": 5304
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.390543507593715e-05,
+      "loss": 1.1451,
+      "step": 5305
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.387127301126371e-05,
+      "loss": 1.3273,
+      "step": 5306
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.383711287905174e-05,
+      "loss": 1.3339,
+      "step": 5307
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.380295468339424e-05,
+      "loss": 1.2033,
+      "step": 5308
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.376879842838381e-05,
+      "loss": 1.1889,
+      "step": 5309
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.373464411811287e-05,
+      "loss": 1.3363,
+      "step": 5310
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.370049175667367e-05,
+      "loss": 1.4522,
+      "step": 5311
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.366634134815812e-05,
+      "loss": 1.5792,
+      "step": 5312
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.363219289665799e-05,
+      "loss": 0.985,
+      "step": 5313
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.359804640626469e-05,
+      "loss": 1.4865,
+      "step": 5314
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.35639018810696e-05,
+      "loss": 1.3308,
+      "step": 5315
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.352975932516364e-05,
+      "loss": 1.3067,
+      "step": 5316
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.349561874263765e-05,
+      "loss": 1.2728,
+      "step": 5317
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.346148013758215e-05,
+      "loss": 1.1138,
+      "step": 5318
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.342734351408748e-05,
+      "loss": 1.2467,
+      "step": 5319
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.339320887624367e-05,
+      "loss": 0.9919,
+      "step": 5320
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.335907622814063e-05,
+      "loss": 0.9299,
+      "step": 5321
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.33249455738679e-05,
+      "loss": 1.4685,
+      "step": 5322
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.329081691751487e-05,
+      "loss": 1.3776,
+      "step": 5323
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.32566902631706e-05,
+      "loss": 1.6528,
+      "step": 5324
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.322256561492408e-05,
+      "loss": 1.2008,
+      "step": 5325
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.318844297686387e-05,
+      "loss": 1.0734,
+      "step": 5326
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.31543223530784e-05,
+      "loss": 1.0042,
+      "step": 5327
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.312020374765581e-05,
+      "loss": 1.2586,
+      "step": 5328
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.308608716468405e-05,
+      "loss": 1.104,
+      "step": 5329
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.305197260825072e-05,
+      "loss": 1.5045,
+      "step": 5330
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.301786008244334e-05,
+      "loss": 1.4015,
+      "step": 5331
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.298374959134905e-05,
+      "loss": 1.0092,
+      "step": 5332
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.294964113905482e-05,
+      "loss": 0.9552,
+      "step": 5333
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.291553472964731e-05,
+      "loss": 1.5222,
+      "step": 5334
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.288143036721303e-05,
+      "loss": 1.5801,
+      "step": 5335
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.284732805583812e-05,
+      "loss": 1.2329,
+      "step": 5336
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.281322779960864e-05,
+      "loss": 1.4174,
+      "step": 5337
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.27791296026102e-05,
+      "loss": 1.3618,
+      "step": 5338
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.274503346892839e-05,
+      "loss": 1.3702,
+      "step": 5339
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.27109394026483e-05,
+      "loss": 1.6784,
+      "step": 5340
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.267684740785502e-05,
+      "loss": 1.0339,
+      "step": 5341
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.264275748863322e-05,
+      "loss": 1.2526,
+      "step": 5342
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.260866964906736e-05,
+      "loss": 0.9987,
+      "step": 5343
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.257458389324175e-05,
+      "loss": 1.4904,
+      "step": 5344
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.254050022524032e-05,
+      "loss": 1.2129,
+      "step": 5345
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.250641864914682e-05,
+      "loss": 1.4053,
+      "step": 5346
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.247233916904466e-05,
+      "loss": 1.178,
+      "step": 5347
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.243826178901718e-05,
+      "loss": 1.456,
+      "step": 5348
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.240418651314727e-05,
+      "loss": 1.144,
+      "step": 5349
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.237011334551773e-05,
+      "loss": 1.3262,
+      "step": 5350
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.233604229021093e-05,
+      "loss": 1.5001,
+      "step": 5351
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.230197335130923e-05,
+      "loss": 1.3697,
+      "step": 5352
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.226790653289444e-05,
+      "loss": 1.7067,
+      "step": 5353
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.223384183904839e-05,
+      "loss": 1.2359,
+      "step": 5354
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.219977927385247e-05,
+      "loss": 1.6287,
+      "step": 5355
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.216571884138793e-05,
+      "loss": 1.1846,
+      "step": 5356
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.213166054573565e-05,
+      "loss": 1.5737,
+      "step": 5357
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.209760439097642e-05,
+      "loss": 1.4809,
+      "step": 5358
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.206355038119058e-05,
+      "loss": 1.2529,
+      "step": 5359
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.202949852045838e-05,
+      "loss": 1.2146,
+      "step": 5360
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.199544881285966e-05,
+      "loss": 1.0843,
+      "step": 5361
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.196140126247416e-05,
+      "loss": 1.0864,
+      "step": 5362
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.192735587338124e-05,
+      "loss": 1.1685,
+      "step": 5363
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.189331264966006e-05,
+      "loss": 1.4115,
+      "step": 5364
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.185927159538949e-05,
+      "loss": 1.315,
+      "step": 5365
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.18252327146482e-05,
+      "loss": 0.9648,
+      "step": 5366
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.179119601151446e-05,
+      "loss": 1.1464,
+      "step": 5367
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.175716149006649e-05,
+      "loss": 1.6662,
+      "step": 5368
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.172312915438203e-05,
+      "loss": 1.3278,
+      "step": 5369
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.168909900853874e-05,
+      "loss": 1.2929,
+      "step": 5370
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.165507105661386e-05,
+      "loss": 1.1937,
+      "step": 5371
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.162104530268453e-05,
+      "loss": 0.9502,
+      "step": 5372
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.158702175082747e-05,
+      "loss": 1.0498,
+      "step": 5373
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.155300040511927e-05,
+      "loss": 1.2571,
+      "step": 5374
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.15189812696361e-05,
+      "loss": 1.1328,
+      "step": 5375
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.148496434845411e-05,
+      "loss": 1.1674,
+      "step": 5376
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.145094964564886e-05,
+      "loss": 1.3912,
+      "step": 5377
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.141693716529596e-05,
+      "loss": 1.3802,
+      "step": 5378
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.138292691147053e-05,
+      "loss": 1.2574,
+      "step": 5379
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.13489188882475e-05,
+      "loss": 1.1623,
+      "step": 5380
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.131491309970159e-05,
+      "loss": 1.1602,
+      "step": 5381
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.128090954990717e-05,
+      "loss": 1.4471,
+      "step": 5382
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.124690824293839e-05,
+      "loss": 1.2594,
+      "step": 5383
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.121290918286907e-05,
+      "loss": 1.2195,
+      "step": 5384
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.117891237377286e-05,
+      "loss": 1.2641,
+      "step": 5385
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.114491781972304e-05,
+      "loss": 1.3756,
+      "step": 5386
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.11109255247927e-05,
+      "loss": 1.0367,
+      "step": 5387
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.107693549305456e-05,
+      "loss": 1.4987,
+      "step": 5388
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.104294772858122e-05,
+      "loss": 1.341,
+      "step": 5389
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.10089622354448e-05,
+      "loss": 1.238,
+      "step": 5390
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.097497901771741e-05,
+      "loss": 1.1377,
+      "step": 5391
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.094099807947065e-05,
+      "loss": 1.248,
+      "step": 5392
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.090701942477599e-05,
+      "loss": 1.0644,
+      "step": 5393
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.08730430577045e-05,
+      "loss": 1.0749,
+      "step": 5394
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.083906898232716e-05,
+      "loss": 1.1216,
+      "step": 5395
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.080509720271452e-05,
+      "loss": 1.3718,
+      "step": 5396
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.077112772293691e-05,
+      "loss": 1.3241,
+      "step": 5397
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.073716054706433e-05,
+      "loss": 1.4577,
+      "step": 5398
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.070319567916667e-05,
+      "loss": 1.2391,
+      "step": 5399
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.066923312331328e-05,
+      "loss": 1.2709,
+      "step": 5400
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.06352728835735e-05,
+      "loss": 1.0975,
+      "step": 5401
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.060131496401621e-05,
+      "loss": 1.2148,
+      "step": 5402
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.05673593687101e-05,
+      "loss": 1.2384,
+      "step": 5403
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.053340610172352e-05,
+      "loss": 1.1469,
+      "step": 5404
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.049945516712464e-05,
+      "loss": 0.6871,
+      "step": 5405
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.046550656898122e-05,
+      "loss": 1.0042,
+      "step": 5406
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.043156031136087e-05,
+      "loss": 1.1678,
+      "step": 5407
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.039761639833078e-05,
+      "loss": 1.1512,
+      "step": 5408
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.036367483395802e-05,
+      "loss": 1.7184,
+      "step": 5409
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.032973562230924e-05,
+      "loss": 1.107,
+      "step": 5410
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.02957987674509e-05,
+      "loss": 1.1499,
+      "step": 5411
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.026186427344908e-05,
+      "loss": 0.9731,
+      "step": 5412
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.02279321443697e-05,
+      "loss": 1.0043,
+      "step": 5413
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.019400238427828e-05,
+      "loss": 1.2962,
+      "step": 5414
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.016007499724015e-05,
+      "loss": 1.2217,
+      "step": 5415
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.012614998732033e-05,
+      "loss": 1.3481,
+      "step": 5416
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.009222735858348e-05,
+      "loss": 1.4355,
+      "step": 5417
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.00583071150941e-05,
+      "loss": 1.5951,
+      "step": 5418
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.002438926091631e-05,
+      "loss": 1.1278,
+      "step": 5419
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.999047380011398e-05,
+      "loss": 1.5269,
+      "step": 5420
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.995656073675066e-05,
+      "loss": 1.0854,
+      "step": 5421
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.992265007488968e-05,
+      "loss": 1.3834,
+      "step": 5422
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.988874181859402e-05,
+      "loss": 1.0965,
+      "step": 5423
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.985483597192642e-05,
+      "loss": 0.9289,
+      "step": 5424
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.982093253894925e-05,
+      "loss": 1.2239,
+      "step": 5425
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.978703152372471e-05,
+      "loss": 1.4153,
+      "step": 5426
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.975313293031458e-05,
+      "loss": 1.2321,
+      "step": 5427
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.971923676278048e-05,
+      "loss": 1.1048,
+      "step": 5428
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.968534302518364e-05,
+      "loss": 1.6907,
+      "step": 5429
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.965145172158505e-05,
+      "loss": 1.2849,
+      "step": 5430
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.961756285604533e-05,
+      "loss": 1.386,
+      "step": 5431
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.958367643262499e-05,
+      "loss": 1.1936,
+      "step": 5432
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.954979245538404e-05,
+      "loss": 1.452,
+      "step": 5433
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.951591092838231e-05,
+      "loss": 1.3249,
+      "step": 5434
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.948203185567929e-05,
+      "loss": 1.1916,
+      "step": 5435
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.944815524133425e-05,
+      "loss": 1.14,
+      "step": 5436
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.941428108940603e-05,
+      "loss": 1.1958,
+      "step": 5437
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.938040940395335e-05,
+      "loss": 1.0872,
+      "step": 5438
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.934654018903446e-05,
+      "loss": 1.0256,
+      "step": 5439
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.931267344870746e-05,
+      "loss": 1.307,
+      "step": 5440
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.927880918703004e-05,
+      "loss": 1.0497,
+      "step": 5441
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.924494740805968e-05,
+      "loss": 1.0561,
+      "step": 5442
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.92110881158535e-05,
+      "loss": 1.0011,
+      "step": 5443
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.917723131446838e-05,
+      "loss": 1.3832,
+      "step": 5444
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.91433770079608e-05,
+      "loss": 1.3586,
+      "step": 5445
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.91095252003871e-05,
+      "loss": 1.2208,
+      "step": 5446
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.907567589580316e-05,
+      "loss": 1.1193,
+      "step": 5447
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.90418290982647e-05,
+      "loss": 1.2347,
+      "step": 5448
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.900798481182699e-05,
+      "loss": 1.1615,
+      "step": 5449
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.897414304054514e-05,
+      "loss": 1.3806,
+      "step": 5450
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.894030378847381e-05,
+      "loss": 1.3023,
+      "step": 5451
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.890646705966756e-05,
+      "loss": 1.3275,
+      "step": 5452
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.887263285818048e-05,
+      "loss": 1.1098,
+      "step": 5453
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.883880118806639e-05,
+      "loss": 1.2111,
+      "step": 5454
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.880497205337888e-05,
+      "loss": 1.6155,
+      "step": 5455
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.877114545817115e-05,
+      "loss": 1.4275,
+      "step": 5456
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.873732140649615e-05,
+      "loss": 0.9728,
+      "step": 5457
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.870349990240644e-05,
+      "loss": 1.5828,
+      "step": 5458
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.866968094995447e-05,
+      "loss": 1.3885,
+      "step": 5459
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.86358645531921e-05,
+      "loss": 1.2971,
+      "step": 5460
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.860205071617115e-05,
+      "loss": 1.3745,
+      "step": 5461
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.856823944294296e-05,
+      "loss": 1.2332,
+      "step": 5462
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.853443073755866e-05,
+      "loss": 1.4081,
+      "step": 5463
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.8500624604069e-05,
+      "loss": 1.1411,
+      "step": 5464
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.846682104652451e-05,
+      "loss": 1.1268,
+      "step": 5465
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.84330200689753e-05,
+      "loss": 1.0033,
+      "step": 5466
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.839922167547129e-05,
+      "loss": 1.0815,
+      "step": 5467
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.836542587006197e-05,
+      "loss": 1.0345,
+      "step": 5468
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.833163265679663e-05,
+      "loss": 1.3163,
+      "step": 5469
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.829784203972417e-05,
+      "loss": 1.2282,
+      "step": 5470
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.826405402289322e-05,
+      "loss": 1.3165,
+      "step": 5471
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.823026861035209e-05,
+      "loss": 1.1832,
+      "step": 5472
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.819648580614878e-05,
+      "loss": 1.5189,
+      "step": 5473
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.816270561433092e-05,
+      "loss": 1.4316,
+      "step": 5474
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.812892803894598e-05,
+      "loss": 1.3333,
+      "step": 5475
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.809515308404093e-05,
+      "loss": 1.4036,
+      "step": 5476
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.806138075366256e-05,
+      "loss": 1.4221,
+      "step": 5477
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.802761105185724e-05,
+      "loss": 1.1434,
+      "step": 5478
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.799384398267117e-05,
+      "loss": 1.3534,
+      "step": 5479
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.796007955015007e-05,
+      "loss": 1.4371,
+      "step": 5480
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.792631775833947e-05,
+      "loss": 1.2454,
+      "step": 5481
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.789255861128449e-05,
+      "loss": 1.2286,
+      "step": 5482
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.785880211303006e-05,
+      "loss": 1.2547,
+      "step": 5483
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.78250482676206e-05,
+      "loss": 0.8105,
+      "step": 5484
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.77912970791004e-05,
+      "loss": 1.1759,
+      "step": 5485
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.775754855151334e-05,
+      "loss": 1.2571,
+      "step": 5486
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.772380268890295e-05,
+      "loss": 1.2538,
+      "step": 5487
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.769005949531258e-05,
+      "loss": 1.289,
+      "step": 5488
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.765631897478506e-05,
+      "loss": 1.631,
+      "step": 5489
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.762258113136312e-05,
+      "loss": 1.2645,
+      "step": 5490
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.758884596908895e-05,
+      "loss": 1.3039,
+      "step": 5491
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.75551134920046e-05,
+      "loss": 1.2026,
+      "step": 5492
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.752138370415167e-05,
+      "loss": 1.2206,
+      "step": 5493
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.748765660957152e-05,
+      "loss": 1.0657,
+      "step": 5494
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.745393221230513e-05,
+      "loss": 1.5189,
+      "step": 5495
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.742021051639327e-05,
+      "loss": 1.1977,
+      "step": 5496
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.738649152587616e-05,
+      "loss": 1.7065,
+      "step": 5497
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.735277524479397e-05,
+      "loss": 1.5503,
+      "step": 5498
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.731906167718632e-05,
+      "loss": 1.2167,
+      "step": 5499
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.728535082709267e-05,
+      "loss": 1.3713,
+      "step": 5500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.725164269855198e-05,
+      "loss": 0.9887,
+      "step": 5501
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.72179372956031e-05,
+      "loss": 1.2845,
+      "step": 5502
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.718423462228438e-05,
+      "loss": 0.905,
+      "step": 5503
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.715053468263392e-05,
+      "loss": 1.2684,
+      "step": 5504
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.711683748068942e-05,
+      "loss": 1.4383,
+      "step": 5505
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.708314302048842e-05,
+      "loss": 1.3184,
+      "step": 5506
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.70494513060679e-05,
+      "loss": 1.6425,
+      "step": 5507
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.70157623414647e-05,
+      "loss": 1.3584,
+      "step": 5508
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.698207613071523e-05,
+      "loss": 1.3269,
+      "step": 5509
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.694839267785563e-05,
+      "loss": 1.628,
+      "step": 5510
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.69147119869216e-05,
+      "loss": 1.2513,
+      "step": 5511
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.688103406194868e-05,
+      "loss": 1.7334,
+      "step": 5512
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.684735890697195e-05,
+      "loss": 1.4605,
+      "step": 5513
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.681368652602621e-05,
+      "loss": 1.5444,
+      "step": 5514
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.678001692314585e-05,
+      "loss": 1.5808,
+      "step": 5515
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.674635010236507e-05,
+      "loss": 0.9955,
+      "step": 5516
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.671268606771762e-05,
+      "loss": 1.44,
+      "step": 5517
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.667902482323696e-05,
+      "loss": 1.2904,
+      "step": 5518
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.664536637295619e-05,
+      "loss": 1.5018,
+      "step": 5519
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.661171072090814e-05,
+      "loss": 1.1769,
+      "step": 5520
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.657805787112517e-05,
+      "loss": 1.3371,
+      "step": 5521
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.654440782763949e-05,
+      "loss": 1.2652,
+      "step": 5522
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.65107605944828e-05,
+      "loss": 1.2659,
+      "step": 5523
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.647711617568656e-05,
+      "loss": 1.2148,
+      "step": 5524
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.644347457528193e-05,
+      "loss": 0.7292,
+      "step": 5525
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.640983579729961e-05,
+      "loss": 1.2706,
+      "step": 5526
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.637619984577007e-05,
+      "loss": 1.1807,
+      "step": 5527
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.634256672472332e-05,
+      "loss": 0.9668,
+      "step": 5528
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.630893643818921e-05,
+      "loss": 1.337,
+      "step": 5529
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.627530899019709e-05,
+      "loss": 1.7798,
+      "step": 5530
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.624168438477604e-05,
+      "loss": 1.0849,
+      "step": 5531
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.620806262595477e-05,
+      "loss": 1.2717,
+      "step": 5532
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.61744437177617e-05,
+      "loss": 1.1947,
+      "step": 5533
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.614082766422482e-05,
+      "loss": 1.2769,
+      "step": 5534
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.610721446937192e-05,
+      "loss": 1.0343,
+      "step": 5535
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.607360413723029e-05,
+      "loss": 1.2806,
+      "step": 5536
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.6039996671827e-05,
+      "loss": 1.2263,
+      "step": 5537
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.600639207718863e-05,
+      "loss": 1.5083,
+      "step": 5538
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.597279035734165e-05,
+      "loss": 1.2594,
+      "step": 5539
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.593919151631193e-05,
+      "loss": 1.2149,
+      "step": 5540
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.59055955581252e-05,
+      "loss": 1.1429,
+      "step": 5541
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.587200248680664e-05,
+      "loss": 1.4859,
+      "step": 5542
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.583841230638135e-05,
+      "loss": 1.6376,
+      "step": 5543
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.580482502087378e-05,
+      "loss": 1.3089,
+      "step": 5544
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.577124063430831e-05,
+      "loss": 1.6365,
+      "step": 5545
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.573765915070878e-05,
+      "loss": 1.4104,
+      "step": 5546
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.570408057409879e-05,
+      "loss": 1.6048,
+      "step": 5547
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.56705049085015e-05,
+      "loss": 1.0474,
+      "step": 5548
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.563693215793983e-05,
+      "loss": 1.7252,
+      "step": 5549
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.560336232643627e-05,
+      "loss": 1.2272,
+      "step": 5550
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.5569795418013e-05,
+      "loss": 1.1664,
+      "step": 5551
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.553623143669178e-05,
+      "loss": 1.2737,
+      "step": 5552
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.550267038649417e-05,
+      "loss": 1.0003,
+      "step": 5553
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.546911227144121e-05,
+      "loss": 1.1297,
+      "step": 5554
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.54355570955537e-05,
+      "loss": 1.581,
+      "step": 5555
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.540200486285202e-05,
+      "loss": 1.1281,
+      "step": 5556
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.536845557735625e-05,
+      "loss": 1.26,
+      "step": 5557
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.533490924308606e-05,
+      "loss": 1.4189,
+      "step": 5558
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.530136586406082e-05,
+      "loss": 1.5527,
+      "step": 5559
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.526782544429958e-05,
+      "loss": 1.1801,
+      "step": 5560
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.523428798782087e-05,
+      "loss": 1.3298,
+      "step": 5561
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.520075349864309e-05,
+      "loss": 1.0496,
+      "step": 5562
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.516722198078408e-05,
+      "loss": 1.308,
+      "step": 5563
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.513369343826148e-05,
+      "loss": 1.4172,
+      "step": 5564
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.510016787509246e-05,
+      "loss": 1.1744,
+      "step": 5565
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.506664529529393e-05,
+      "loss": 1.2323,
+      "step": 5566
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.503312570288235e-05,
+      "loss": 1.3059,
+      "step": 5567
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.49996091018739e-05,
+      "loss": 1.4379,
+      "step": 5568
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.496609549628433e-05,
+      "loss": 1.5377,
+      "step": 5569
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.493258489012911e-05,
+      "loss": 1.4591,
+      "step": 5570
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.489907728742325e-05,
+      "loss": 1.4614,
+      "step": 5571
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.486557269218153e-05,
+      "loss": 1.0608,
+      "step": 5572
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.483207110841826e-05,
+      "loss": 1.3437,
+      "step": 5573
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.479857254014746e-05,
+      "loss": 1.1041,
+      "step": 5574
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.476507699138269e-05,
+      "loss": 1.5344,
+      "step": 5575
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.473158446613729e-05,
+      "loss": 1.2635,
+      "step": 5576
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.46980949684241e-05,
+      "loss": 1.6384,
+      "step": 5577
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.466460850225573e-05,
+      "loss": 1.2,
+      "step": 5578
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.463112507164431e-05,
+      "loss": 1.3627,
+      "step": 5579
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.459764468060166e-05,
+      "loss": 1.3233,
+      "step": 5580
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.45641673331392e-05,
+      "loss": 1.1728,
+      "step": 5581
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.453069303326807e-05,
+      "loss": 1.5422,
+      "step": 5582
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.449722178499896e-05,
+      "loss": 1.1387,
+      "step": 5583
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.446375359234225e-05,
+      "loss": 1.5161,
+      "step": 5584
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.443028845930786e-05,
+      "loss": 1.1907,
+      "step": 5585
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.43968263899055e-05,
+      "loss": 1.3776,
+      "step": 5586
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.436336738814437e-05,
+      "loss": 1.1289,
+      "step": 5587
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.432991145803337e-05,
+      "loss": 1.0584,
+      "step": 5588
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.429645860358098e-05,
+      "loss": 1.4951,
+      "step": 5589
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.426300882879546e-05,
+      "loss": 1.3352,
+      "step": 5590
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.422956213768443e-05,
+      "loss": 1.5234,
+      "step": 5591
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.419611853425546e-05,
+      "loss": 1.2866,
+      "step": 5592
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.416267802251546e-05,
+      "loss": 1.4973,
+      "step": 5593
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.412924060647122e-05,
+      "loss": 1.1958,
+      "step": 5594
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.409580629012892e-05,
+      "loss": 1.4924,
+      "step": 5595
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.406237507749456e-05,
+      "loss": 1.0259,
+      "step": 5596
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.402894697257372e-05,
+      "loss": 1.1527,
+      "step": 5597
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.39955219793715e-05,
+      "loss": 1.6697,
+      "step": 5598
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.396210010189282e-05,
+      "loss": 1.4807,
+      "step": 5599
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.392868134414201e-05,
+      "loss": 1.2738,
+      "step": 5600
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.389526571012323e-05,
+      "loss": 1.4543,
+      "step": 5601
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.386185320384007e-05,
+      "loss": 1.2968,
+      "step": 5602
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.382844382929597e-05,
+      "loss": 1.4835,
+      "step": 5603
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.379503759049373e-05,
+      "loss": 1.1903,
+      "step": 5604
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.376163449143605e-05,
+      "loss": 1.1316,
+      "step": 5605
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.3728234536125e-05,
+      "loss": 1.2895,
+      "step": 5606
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.36948377285625e-05,
+      "loss": 1.0453,
+      "step": 5607
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.366144407274986e-05,
+      "loss": 1.1543,
+      "step": 5608
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.362805357268826e-05,
+      "loss": 1.1806,
+      "step": 5609
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.35946662323783e-05,
+      "loss": 1.1298,
+      "step": 5610
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.356128205582033e-05,
+      "loss": 1.4011,
+      "step": 5611
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.352790104701421e-05,
+      "loss": 1.5008,
+      "step": 5612
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.349452320995955e-05,
+      "loss": 1.0514,
+      "step": 5613
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.346114854865547e-05,
+      "loss": 1.3014,
+      "step": 5614
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.342777706710078e-05,
+      "loss": 1.0345,
+      "step": 5615
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.339440876929384e-05,
+      "loss": 1.3122,
+      "step": 5616
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.336104365923271e-05,
+      "loss": 1.205,
+      "step": 5617
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.332768174091497e-05,
+      "loss": 1.3429,
+      "step": 5618
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.329432301833794e-05,
+      "loss": 1.0884,
+      "step": 5619
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.326096749549846e-05,
+      "loss": 1.3148,
+      "step": 5620
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.322761517639303e-05,
+      "loss": 1.3669,
+      "step": 5621
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.31942660650177e-05,
+      "loss": 1.3566,
+      "step": 5622
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.316092016536828e-05,
+      "loss": 1.2365,
+      "step": 5623
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.312757748144006e-05,
+      "loss": 1.5854,
+      "step": 5624
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.309423801722799e-05,
+      "loss": 1.3122,
+      "step": 5625
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.306090177672658e-05,
+      "loss": 1.3981,
+      "step": 5626
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.302756876393015e-05,
+      "loss": 1.3054,
+      "step": 5627
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.299423898283233e-05,
+      "loss": 1.3026,
+      "step": 5628
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.296091243742663e-05,
+      "loss": 1.3952,
+      "step": 5629
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.292758913170603e-05,
+      "loss": 1.4351,
+      "step": 5630
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.289426906966313e-05,
+      "loss": 1.6193,
+      "step": 5631
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.286095225529024e-05,
+      "loss": 1.41,
+      "step": 5632
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.282763869257913e-05,
+      "loss": 1.7378,
+      "step": 5633
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.279432838552135e-05,
+      "loss": 1.3095,
+      "step": 5634
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.276102133810787e-05,
+      "loss": 1.4439,
+      "step": 5635
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.272771755432946e-05,
+      "loss": 1.2162,
+      "step": 5636
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.269441703817635e-05,
+      "loss": 1.4208,
+      "step": 5637
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.266111979363848e-05,
+      "loss": 1.3711,
+      "step": 5638
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.262782582470533e-05,
+      "loss": 1.6339,
+      "step": 5639
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.259453513536601e-05,
+      "loss": 1.4344,
+      "step": 5640
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.256124772960922e-05,
+      "loss": 1.1554,
+      "step": 5641
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.252796361142334e-05,
+      "loss": 1.3296,
+      "step": 5642
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.249468278479626e-05,
+      "loss": 1.2601,
+      "step": 5643
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.246140525371556e-05,
+      "loss": 1.395,
+      "step": 5644
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.242813102216831e-05,
+      "loss": 1.2692,
+      "step": 5645
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.239486009414135e-05,
+      "loss": 1.4341,
+      "step": 5646
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.236159247362097e-05,
+      "loss": 1.5491,
+      "step": 5647
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.232832816459316e-05,
+      "loss": 1.282,
+      "step": 5648
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.229506717104342e-05,
+      "loss": 1.5947,
+      "step": 5649
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.226180949695702e-05,
+      "loss": 1.2887,
+      "step": 5650
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.222855514631859e-05,
+      "loss": 1.1242,
+      "step": 5651
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.219530412311262e-05,
+      "loss": 1.2461,
+      "step": 5652
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.216205643132298e-05,
+      "loss": 1.2568,
+      "step": 5653
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.212881207493331e-05,
+      "loss": 1.1598,
+      "step": 5654
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.209557105792673e-05,
+      "loss": 1.7191,
+      "step": 5655
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.206233338428605e-05,
+      "loss": 1.3414,
+      "step": 5656
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.20290990579936e-05,
+      "loss": 1.5505,
+      "step": 5657
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.199586808303138e-05,
+      "loss": 1.5236,
+      "step": 5658
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.19626404633809e-05,
+      "loss": 1.3535,
+      "step": 5659
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.192941620302342e-05,
+      "loss": 1.2736,
+      "step": 5660
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.189619530593961e-05,
+      "loss": 1.0864,
+      "step": 5661
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.186297777610986e-05,
+      "loss": 1.4353,
+      "step": 5662
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.182976361751412e-05,
+      "loss": 1.3017,
+      "step": 5663
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.179655283413198e-05,
+      "loss": 1.5523,
+      "step": 5664
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.17633454299425e-05,
+      "loss": 1.7666,
+      "step": 5665
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.17301414089245e-05,
+      "loss": 1.4511,
+      "step": 5666
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.169694077505629e-05,
+      "loss": 1.5157,
+      "step": 5667
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.166374353231578e-05,
+      "loss": 1.5756,
+      "step": 5668
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.163054968468053e-05,
+      "loss": 1.0309,
+      "step": 5669
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.159735923612763e-05,
+      "loss": 1.5287,
+      "step": 5670
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.156417219063383e-05,
+      "loss": 0.9186,
+      "step": 5671
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.153098855217535e-05,
+      "loss": 1.0538,
+      "step": 5672
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.14978083247282e-05,
+      "loss": 1.4003,
+      "step": 5673
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.146463151226774e-05,
+      "loss": 0.9429,
+      "step": 5674
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.143145811876914e-05,
+      "loss": 1.216,
+      "step": 5675
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.139828814820702e-05,
+      "loss": 1.139,
+      "step": 5676
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.136512160455567e-05,
+      "loss": 1.3666,
+      "step": 5677
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.133195849178887e-05,
+      "loss": 1.3764,
+      "step": 5678
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.129879881388014e-05,
+      "loss": 1.2819,
+      "step": 5679
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.126564257480243e-05,
+      "loss": 1.0659,
+      "step": 5680
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.123248977852841e-05,
+      "loss": 1.173,
+      "step": 5681
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.11993404290302e-05,
+      "loss": 1.4325,
+      "step": 5682
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.116619453027969e-05,
+      "loss": 1.3051,
+      "step": 5683
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.113305208624818e-05,
+      "loss": 1.2417,
+      "step": 5684
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.109991310090666e-05,
+      "loss": 1.3741,
+      "step": 5685
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.106677757822562e-05,
+      "loss": 1.1597,
+      "step": 5686
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.103364552217529e-05,
+      "loss": 1.2427,
+      "step": 5687
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.100051693672524e-05,
+      "loss": 0.9589,
+      "step": 5688
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.096739182584494e-05,
+      "loss": 1.5255,
+      "step": 5689
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.093427019350312e-05,
+      "loss": 0.9842,
+      "step": 5690
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.090115204366832e-05,
+      "loss": 1.3581,
+      "step": 5691
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.086803738030857e-05,
+      "loss": 1.4442,
+      "step": 5692
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.083492620739151e-05,
+      "loss": 0.8983,
+      "step": 5693
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.080181852888432e-05,
+      "loss": 1.1038,
+      "step": 5694
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.076871434875385e-05,
+      "loss": 1.2766,
+      "step": 5695
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.07356136709664e-05,
+      "loss": 1.064,
+      "step": 5696
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.0702516499488e-05,
+      "loss": 0.8371,
+      "step": 5697
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.066942283828413e-05,
+      "loss": 1.3268,
+      "step": 5698
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.063633269131993e-05,
+      "loss": 1.397,
+      "step": 5699
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.060324606256009e-05,
+      "loss": 1.0384,
+      "step": 5700
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.057016295596885e-05,
+      "loss": 1.2021,
+      "step": 5701
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.053708337551006e-05,
+      "loss": 1.3607,
+      "step": 5702
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.050400732514719e-05,
+      "loss": 1.321,
+      "step": 5703
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.047093480884324e-05,
+      "loss": 1.3256,
+      "step": 5704
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.043786583056073e-05,
+      "loss": 1.1718,
+      "step": 5705
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.04048003942619e-05,
+      "loss": 1.5194,
+      "step": 5706
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.037173850390841e-05,
+      "loss": 1.4383,
+      "step": 5707
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.033868016346161e-05,
+      "loss": 0.8842,
+      "step": 5708
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.030562537688233e-05,
+      "loss": 1.2402,
+      "step": 5709
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.027257414813113e-05,
+      "loss": 1.3793,
+      "step": 5710
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.02395264811679e-05,
+      "loss": 0.9179,
+      "step": 5711
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.020648237995235e-05,
+      "loss": 1.4183,
+      "step": 5712
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.017344184844358e-05,
+      "loss": 1.2769,
+      "step": 5713
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.014040489060042e-05,
+      "loss": 1.7093,
+      "step": 5714
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.010737151038109e-05,
+      "loss": 1.4473,
+      "step": 5715
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.007434171174358e-05,
+      "loss": 1.1168,
+      "step": 5716
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.004131549864528e-05,
+      "loss": 0.9072,
+      "step": 5717
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.000829287504327e-05,
+      "loss": 1.3981,
+      "step": 5718
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.997527384489409e-05,
+      "loss": 1.5486,
+      "step": 5719
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.994225841215398e-05,
+      "loss": 1.7119,
+      "step": 5720
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.990924658077865e-05,
+      "loss": 1.3515,
+      "step": 5721
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.987623835472342e-05,
+      "loss": 1.2854,
+      "step": 5722
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.984323373794312e-05,
+      "loss": 1.4645,
+      "step": 5723
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.981023273439226e-05,
+      "loss": 1.3682,
+      "step": 5724
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.977723534802481e-05,
+      "loss": 1.1135,
+      "step": 5725
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.974424158279435e-05,
+      "loss": 1.3723,
+      "step": 5726
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.971125144265405e-05,
+      "loss": 1.4115,
+      "step": 5727
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.96782649315566e-05,
+      "loss": 1.5103,
+      "step": 5728
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.964528205345427e-05,
+      "loss": 1.2853,
+      "step": 5729
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.961230281229891e-05,
+      "loss": 1.4049,
+      "step": 5730
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.957932721204192e-05,
+      "loss": 1.2248,
+      "step": 5731
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.954635525663429e-05,
+      "loss": 1.0763,
+      "step": 5732
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.951338695002649e-05,
+      "loss": 1.2944,
+      "step": 5733
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.948042229616872e-05,
+      "loss": 1.4199,
+      "step": 5734
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.944746129901049e-05,
+      "loss": 1.13,
+      "step": 5735
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.941450396250115e-05,
+      "loss": 1.1434,
+      "step": 5736
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.938155029058941e-05,
+      "loss": 1.6212,
+      "step": 5737
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.934860028722366e-05,
+      "loss": 1.249,
+      "step": 5738
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.93156539563517e-05,
+      "loss": 1.4177,
+      "step": 5739
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.92827113019211e-05,
+      "loss": 1.4054,
+      "step": 5740
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.924977232787884e-05,
+      "loss": 1.5905,
+      "step": 5741
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.921683703817148e-05,
+      "loss": 1.3034,
+      "step": 5742
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.918390543674521e-05,
+      "loss": 1.2705,
+      "step": 5743
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.915097752754567e-05,
+      "loss": 1.3096,
+      "step": 5744
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.911805331451816e-05,
+      "loss": 1.4349,
+      "step": 5745
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.908513280160742e-05,
+      "loss": 0.9689,
+      "step": 5746
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.905221599275794e-05,
+      "loss": 1.2654,
+      "step": 5747
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.90193028919135e-05,
+      "loss": 1.2763,
+      "step": 5748
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.89863935030177e-05,
+      "loss": 1.1208,
+      "step": 5749
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.895348783001347e-05,
+      "loss": 1.5476,
+      "step": 5750
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.89205858768435e-05,
+      "loss": 1.2318,
+      "step": 5751
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.888768764744984e-05,
+      "loss": 1.5814,
+      "step": 5752
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.885479314577427e-05,
+      "loss": 1.159,
+      "step": 5753
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.882190237575798e-05,
+      "loss": 1.2047,
+      "step": 5754
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.878901534134181e-05,
+      "loss": 1.1584,
+      "step": 5755
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.875613204646608e-05,
+      "loss": 0.941,
+      "step": 5756
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.872325249507075e-05,
+      "loss": 1.1903,
+      "step": 5757
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.869037669109524e-05,
+      "loss": 1.0523,
+      "step": 5758
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.865750463847859e-05,
+      "loss": 0.983,
+      "step": 5759
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.862463634115932e-05,
+      "loss": 1.2552,
+      "step": 5760
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.859177180307557e-05,
+      "loss": 1.149,
+      "step": 5761
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.855891102816498e-05,
+      "loss": 1.2921,
+      "step": 5762
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.85260540203648e-05,
+      "loss": 1.1098,
+      "step": 5763
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.849320078361174e-05,
+      "loss": 1.2414,
+      "step": 5764
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.846035132184216e-05,
+      "loss": 1.2502,
+      "step": 5765
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.842750563899185e-05,
+      "loss": 1.3073,
+      "step": 5766
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.839466373899627e-05,
+      "loss": 1.2071,
+      "step": 5767
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.836182562579031e-05,
+      "loss": 1.2118,
+      "step": 5768
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.832899130330856e-05,
+      "loss": 1.6522,
+      "step": 5769
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.829616077548495e-05,
+      "loss": 1.1784,
+      "step": 5770
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.826333404625316e-05,
+      "loss": 1.2227,
+      "step": 5771
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.823051111954621e-05,
+      "loss": 1.3252,
+      "step": 5772
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.819769199929689e-05,
+      "loss": 1.4378,
+      "step": 5773
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.816487668943736e-05,
+      "loss": 1.4601,
+      "step": 5774
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.813206519389937e-05,
+      "loss": 1.1704,
+      "step": 5775
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.809925751661427e-05,
+      "loss": 1.3639,
+      "step": 5776
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.806645366151289e-05,
+      "loss": 1.3077,
+      "step": 5777
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.803365363252562e-05,
+      "loss": 1.4244,
+      "step": 5778
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.800085743358236e-05,
+      "loss": 1.1519,
+      "step": 5779
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.796806506861265e-05,
+      "loss": 0.9192,
+      "step": 5780
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.793527654154542e-05,
+      "loss": 1.1367,
+      "step": 5781
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.79024918563093e-05,
+      "loss": 1.0891,
+      "step": 5782
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.786971101683235e-05,
+      "loss": 1.3385,
+      "step": 5783
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.783693402704221e-05,
+      "loss": 1.412,
+      "step": 5784
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.7804160890866e-05,
+      "loss": 1.2574,
+      "step": 5785
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.777139161223052e-05,
+      "loss": 1.1625,
+      "step": 5786
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.773862619506194e-05,
+      "loss": 1.1222,
+      "step": 5787
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.77058646432861e-05,
+      "loss": 1.122,
+      "step": 5788
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.767310696082827e-05,
+      "loss": 1.2673,
+      "step": 5789
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.764035315161336e-05,
+      "loss": 1.2609,
+      "step": 5790
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.760760321956573e-05,
+      "loss": 1.2892,
+      "step": 5791
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.757485716860934e-05,
+      "loss": 1.2462,
+      "step": 5792
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.754211500266758e-05,
+      "loss": 1.4618,
+      "step": 5793
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.750937672566357e-05,
+      "loss": 1.478,
+      "step": 5794
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.74766423415197e-05,
+      "loss": 1.38,
+      "step": 5795
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.744391185415816e-05,
+      "loss": 1.051,
+      "step": 5796
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.741118526750049e-05,
+      "loss": 1.1806,
+      "step": 5797
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.737846258546783e-05,
+      "loss": 1.2807,
+      "step": 5798
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.734574381198083e-05,
+      "loss": 1.5575,
+      "step": 5799
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.731302895095972e-05,
+      "loss": 1.1491,
+      "step": 5800
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.72803180063242e-05,
+      "loss": 1.3096,
+      "step": 5801
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.724761098199357e-05,
+      "loss": 1.3297,
+      "step": 5802
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.721490788188653e-05,
+      "loss": 1.0766,
+      "step": 5803
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.718220870992149e-05,
+      "loss": 1.1433,
+      "step": 5804
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.714951347001625e-05,
+      "loss": 1.2693,
+      "step": 5805
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.711682216608822e-05,
+      "loss": 1.4515,
+      "step": 5806
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.708413480205427e-05,
+      "loss": 1.1622,
+      "step": 5807
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.705145138183084e-05,
+      "loss": 1.1232,
+      "step": 5808
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.701877190933387e-05,
+      "loss": 1.3621,
+      "step": 5809
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.698609638847893e-05,
+      "loss": 0.9306,
+      "step": 5810
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.695342482318093e-05,
+      "loss": 1.0998,
+      "step": 5811
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.692075721735444e-05,
+      "loss": 0.9927,
+      "step": 5812
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.688809357491359e-05,
+      "loss": 1.3023,
+      "step": 5813
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.685543389977189e-05,
+      "loss": 1.4508,
+      "step": 5814
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.68227781958425e-05,
+      "loss": 1.3615,
+      "step": 5815
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.679012646703804e-05,
+      "loss": 1.4365,
+      "step": 5816
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.67574787172707e-05,
+      "loss": 1.1599,
+      "step": 5817
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.672483495045211e-05,
+      "loss": 1.3925,
+      "step": 5818
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.669219517049355e-05,
+      "loss": 1.3813,
+      "step": 5819
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.665955938130571e-05,
+      "loss": 1.3158,
+      "step": 5820
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.662692758679884e-05,
+      "loss": 1.1859,
+      "step": 5821
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.65942997908827e-05,
+      "loss": 1.464,
+      "step": 5822
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.656167599746665e-05,
+      "loss": 1.3877,
+      "step": 5823
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.652905621045944e-05,
+      "loss": 1.212,
+      "step": 5824
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.649644043376946e-05,
+      "loss": 1.3305,
+      "step": 5825
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.646382867130449e-05,
+      "loss": 1.4569,
+      "step": 5826
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.643122092697198e-05,
+      "loss": 1.4312,
+      "step": 5827
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.639861720467879e-05,
+      "loss": 1.2751,
+      "step": 5828
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.636601750833135e-05,
+      "loss": 1.1271,
+      "step": 5829
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.633342184183557e-05,
+      "loss": 1.5199,
+      "step": 5830
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.63008302090969e-05,
+      "loss": 0.9099,
+      "step": 5831
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.626824261402027e-05,
+      "loss": 1.2873,
+      "step": 5832
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.623565906051023e-05,
+      "loss": 0.9948,
+      "step": 5833
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.620307955247072e-05,
+      "loss": 1.6219,
+      "step": 5834
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.61705040938053e-05,
+      "loss": 1.5107,
+      "step": 5835
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.613793268841689e-05,
+      "loss": 1.0849,
+      "step": 5836
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.610536534020816e-05,
+      "loss": 1.1839,
+      "step": 5837
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.607280205308108e-05,
+      "loss": 1.5204,
+      "step": 5838
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.604024283093725e-05,
+      "loss": 1.1382,
+      "step": 5839
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.600768767767771e-05,
+      "loss": 1.3066,
+      "step": 5840
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.597513659720314e-05,
+      "loss": 1.1165,
+      "step": 5841
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.594258959341353e-05,
+      "loss": 1.2541,
+      "step": 5842
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.591004667020859e-05,
+      "loss": 1.5484,
+      "step": 5843
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.587750783148739e-05,
+      "loss": 1.0219,
+      "step": 5844
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.584497308114859e-05,
+      "loss": 1.2333,
+      "step": 5845
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.581244242309028e-05,
+      "loss": 1.2144,
+      "step": 5846
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.577991586121022e-05,
+      "loss": 1.1101,
+      "step": 5847
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.574739339940552e-05,
+      "loss": 1.3005,
+      "step": 5848
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.571487504157283e-05,
+      "loss": 1.2901,
+      "step": 5849
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.56823607916084e-05,
+      "loss": 1.3941,
+      "step": 5850
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.564985065340787e-05,
+      "loss": 1.2651,
+      "step": 5851
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.561734463086647e-05,
+      "loss": 1.0868,
+      "step": 5852
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.558484272787884e-05,
+      "loss": 1.2688,
+      "step": 5853
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.55523449483393e-05,
+      "loss": 1.5884,
+      "step": 5854
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.551985129614144e-05,
+      "loss": 1.2825,
+      "step": 5855
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.54873617751786e-05,
+      "loss": 1.0306,
+      "step": 5856
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.545487638934345e-05,
+      "loss": 1.3772,
+      "step": 5857
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.542239514252823e-05,
+      "loss": 0.9875,
+      "step": 5858
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.538991803862465e-05,
+      "loss": 1.0087,
+      "step": 5859
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.535744508152401e-05,
+      "loss": 1.3973,
+      "step": 5860
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.5324976275117e-05,
+      "loss": 1.0751,
+      "step": 5861
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.529251162329391e-05,
+      "loss": 1.1746,
+      "step": 5862
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.526005112994442e-05,
+      "loss": 1.2906,
+      "step": 5863
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.522759479895788e-05,
+      "loss": 1.0798,
+      "step": 5864
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.519514263422296e-05,
+      "loss": 1.4142,
+      "step": 5865
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.516269463962796e-05,
+      "loss": 0.9241,
+      "step": 5866
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.513025081906058e-05,
+      "loss": 0.9659,
+      "step": 5867
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.509781117640817e-05,
+      "loss": 1.1388,
+      "step": 5868
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.506537571555733e-05,
+      "loss": 1.122,
+      "step": 5869
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.503294444039444e-05,
+      "loss": 1.3615,
+      "step": 5870
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.500051735480522e-05,
+      "loss": 1.2753,
+      "step": 5871
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.49680944626749e-05,
+      "loss": 1.0964,
+      "step": 5872
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.49356757678882e-05,
+      "loss": 1.4974,
+      "step": 5873
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.490326127432941e-05,
+      "loss": 1.1724,
+      "step": 5874
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.487085098588226e-05,
+      "loss": 1.5059,
+      "step": 5875
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.483844490642998e-05,
+      "loss": 1.2469,
+      "step": 5876
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.480604303985527e-05,
+      "loss": 1.1862,
+      "step": 5877
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.477364539004044e-05,
+      "loss": 1.2401,
+      "step": 5878
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.474125196086708e-05,
+      "loss": 1.056,
+      "step": 5879
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.470886275621653e-05,
+      "loss": 1.5082,
+      "step": 5880
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.467647777996941e-05,
+      "loss": 1.0076,
+      "step": 5881
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.464409703600601e-05,
+      "loss": 1.0412,
+      "step": 5882
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.46117205282059e-05,
+      "loss": 1.0397,
+      "step": 5883
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.457934826044837e-05,
+      "loss": 1.3571,
+      "step": 5884
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.45469802366121e-05,
+      "loss": 1.1166,
+      "step": 5885
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.451461646057517e-05,
+      "loss": 1.2606,
+      "step": 5886
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.448225693621537e-05,
+      "loss": 1.2739,
+      "step": 5887
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.444990166740977e-05,
+      "loss": 1.2533,
+      "step": 5888
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.441755065803502e-05,
+      "loss": 1.1074,
+      "step": 5889
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.438520391196725e-05,
+      "loss": 1.2885,
+      "step": 5890
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.435286143308212e-05,
+      "loss": 1.4325,
+      "step": 5891
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.432052322525467e-05,
+      "loss": 1.3171,
+      "step": 5892
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.42881892923596e-05,
+      "loss": 1.179,
+      "step": 5893
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.425585963827089e-05,
+      "loss": 0.9916,
+      "step": 5894
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.42235342668622e-05,
+      "loss": 0.7874,
+      "step": 5895
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.41912131820065e-05,
+      "loss": 1.8552,
+      "step": 5896
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.415889638757643e-05,
+      "loss": 1.1881,
+      "step": 5897
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.412658388744397e-05,
+      "loss": 1.5676,
+      "step": 5898
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.409427568548067e-05,
+      "loss": 1.2879,
+      "step": 5899
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.406197178555747e-05,
+      "loss": 0.925,
+      "step": 5900
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.402967219154496e-05,
+      "loss": 1.356,
+      "step": 5901
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3997376907313e-05,
+      "loss": 1.2394,
+      "step": 5902
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.396508593673114e-05,
+      "loss": 1.042,
+      "step": 5903
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.393279928366826e-05,
+      "loss": 1.3344,
+      "step": 5904
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.390051695199282e-05,
+      "loss": 0.6999,
+      "step": 5905
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.386823894557266e-05,
+      "loss": 1.0018,
+      "step": 5906
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.383596526827525e-05,
+      "loss": 1.3747,
+      "step": 5907
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.38036959239674e-05,
+      "loss": 1.4957,
+      "step": 5908
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.377143091651549e-05,
+      "loss": 0.9985,
+      "step": 5909
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.37391702497853e-05,
+      "loss": 1.5709,
+      "step": 5910
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.370691392764221e-05,
+      "loss": 1.3007,
+      "step": 5911
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.367466195395094e-05,
+      "loss": 1.2881,
+      "step": 5912
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.36424143325758e-05,
+      "loss": 1.1158,
+      "step": 5913
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.361017106738053e-05,
+      "loss": 1.5127,
+      "step": 5914
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.357793216222831e-05,
+      "loss": 1.3516,
+      "step": 5915
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.354569762098189e-05,
+      "loss": 1.2787,
+      "step": 5916
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.351346744750342e-05,
+      "loss": 1.0315,
+      "step": 5917
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.348124164565456e-05,
+      "loss": 1.2911,
+      "step": 5918
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.344902021929643e-05,
+      "loss": 1.3984,
+      "step": 5919
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.341680317228971e-05,
+      "loss": 1.08,
+      "step": 5920
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.338459050849439e-05,
+      "loss": 1.2346,
+      "step": 5921
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.335238223177007e-05,
+      "loss": 1.4925,
+      "step": 5922
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.332017834597575e-05,
+      "loss": 1.4251,
+      "step": 5923
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.328797885496997e-05,
+      "loss": 1.5234,
+      "step": 5924
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.32557837626107e-05,
+      "loss": 1.0734,
+      "step": 5925
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.32235930727554e-05,
+      "loss": 1.111,
+      "step": 5926
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.319140678926095e-05,
+      "loss": 1.7046,
+      "step": 5927
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.315922491598381e-05,
+      "loss": 1.1247,
+      "step": 5928
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.312704745677975e-05,
+      "loss": 0.9956,
+      "step": 5929
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.309487441550424e-05,
+      "loss": 1.6207,
+      "step": 5930
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.306270579601197e-05,
+      "loss": 1.3709,
+      "step": 5931
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.303054160215731e-05,
+      "loss": 1.2302,
+      "step": 5932
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.299838183779392e-05,
+      "loss": 1.2585,
+      "step": 5933
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.296622650677511e-05,
+      "loss": 1.2492,
+      "step": 5934
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.29340756129535e-05,
+      "loss": 1.5536,
+      "step": 5935
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.290192916018129e-05,
+      "loss": 1.4246,
+      "step": 5936
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.286978715231005e-05,
+      "loss": 1.0941,
+      "step": 5937
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.283764959319096e-05,
+      "loss": 1.0628,
+      "step": 5938
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.280551648667445e-05,
+      "loss": 1.1876,
+      "step": 5939
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.277338783661068e-05,
+      "loss": 1.5903,
+      "step": 5940
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.274126364684905e-05,
+      "loss": 1.1361,
+      "step": 5941
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.270914392123855e-05,
+      "loss": 1.3418,
+      "step": 5942
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.267702866362755e-05,
+      "loss": 1.3962,
+      "step": 5943
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.264491787786404e-05,
+      "loss": 1.1839,
+      "step": 5944
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.261281156779528e-05,
+      "loss": 1.413,
+      "step": 5945
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.258070973726813e-05,
+      "loss": 1.0317,
+      "step": 5946
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.254861239012881e-05,
+      "loss": 1.2018,
+      "step": 5947
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.251651953022317e-05,
+      "loss": 1.3215,
+      "step": 5948
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.248443116139628e-05,
+      "loss": 1.4027,
+      "step": 5949
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.24523472874929e-05,
+      "loss": 1.3459,
+      "step": 5950
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.242026791235711e-05,
+      "loss": 1.2136,
+      "step": 5951
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.238819303983253e-05,
+      "loss": 1.2357,
+      "step": 5952
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.235612267376214e-05,
+      "loss": 1.2601,
+      "step": 5953
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.232405681798854e-05,
+      "loss": 1.3335,
+      "step": 5954
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.229199547635363e-05,
+      "loss": 1.1003,
+      "step": 5955
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.225993865269884e-05,
+      "loss": 1.3319,
+      "step": 5956
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.22278863508651e-05,
+      "loss": 1.0033,
+      "step": 5957
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.21958385746927e-05,
+      "loss": 1.2189,
+      "step": 5958
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.21637953280215e-05,
+      "loss": 1.1015,
+      "step": 5959
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.213175661469066e-05,
+      "loss": 1.3238,
+      "step": 5960
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.209972243853903e-05,
+      "loss": 1.2375,
+      "step": 5961
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.206769280340464e-05,
+      "loss": 1.6225,
+      "step": 5962
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.203566771312524e-05,
+      "loss": 1.2323,
+      "step": 5963
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.200364717153782e-05,
+      "loss": 1.2933,
+      "step": 5964
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.197163118247897e-05,
+      "loss": 1.3436,
+      "step": 5965
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.193961974978464e-05,
+      "loss": 1.22,
+      "step": 5966
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.190761287729034e-05,
+      "loss": 1.4435,
+      "step": 5967
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.187561056883091e-05,
+      "loss": 1.2356,
+      "step": 5968
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.184361282824073e-05,
+      "loss": 1.1293,
+      "step": 5969
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.18116196593536e-05,
+      "loss": 0.8187,
+      "step": 5970
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.177963106600279e-05,
+      "loss": 1.3116,
+      "step": 5971
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.174764705202098e-05,
+      "loss": 1.0271,
+      "step": 5972
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.171566762124037e-05,
+      "loss": 1.5127,
+      "step": 5973
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.168369277749253e-05,
+      "loss": 0.9224,
+      "step": 5974
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.165172252460855e-05,
+      "loss": 1.2567,
+      "step": 5975
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.161975686641891e-05,
+      "loss": 1.6675,
+      "step": 5976
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.158779580675363e-05,
+      "loss": 1.2696,
+      "step": 5977
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.155583934944207e-05,
+      "loss": 1.2322,
+      "step": 5978
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.152388749831312e-05,
+      "loss": 1.1453,
+      "step": 5979
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.149194025719505e-05,
+      "loss": 1.3773,
+      "step": 5980
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.145999762991566e-05,
+      "loss": 1.4876,
+      "step": 5981
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.142805962030211e-05,
+      "loss": 1.4062,
+      "step": 5982
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.13961262321811e-05,
+      "loss": 1.1233,
+      "step": 5983
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.136419746937865e-05,
+      "loss": 1.188,
+      "step": 5984
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.13322733357204e-05,
+      "loss": 1.5237,
+      "step": 5985
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.130035383503123e-05,
+      "loss": 1.2219,
+      "step": 5986
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.126843897113566e-05,
+      "loss": 1.5877,
+      "step": 5987
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.12365287478575e-05,
+      "loss": 0.9985,
+      "step": 5988
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.120462316902014e-05,
+      "loss": 1.355,
+      "step": 5989
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.117272223844624e-05,
+      "loss": 1.2741,
+      "step": 5990
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.114082595995809e-05,
+      "loss": 1.3184,
+      "step": 5991
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.110893433737735e-05,
+      "loss": 1.5186,
+      "step": 5992
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.107704737452502e-05,
+      "loss": 1.1488,
+      "step": 5993
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.104516507522174e-05,
+      "loss": 1.1195,
+      "step": 5994
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.1013287443287414e-05,
+      "loss": 1.494,
+      "step": 5995
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.09814144825415e-05,
+      "loss": 1.5059,
+      "step": 5996
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0949546196802775e-05,
+      "loss": 1.5402,
+      "step": 5997
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.091768258988967e-05,
+      "loss": 1.1407,
+      "step": 5998
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.088582366561977e-05,
+      "loss": 1.3186,
+      "step": 5999
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.085396942781034e-05,
+      "loss": 1.4556,
+      "step": 6000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0822119880277974e-05,
+      "loss": 1.148,
+      "step": 6001
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.079027502683872e-05,
+      "loss": 1.0274,
+      "step": 6002
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0758434871308014e-05,
+      "loss": 1.7764,
+      "step": 6003
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0726599417500886e-05,
+      "loss": 1.2408,
+      "step": 6004
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0694768669231603e-05,
+      "loss": 1.2598,
+      "step": 6005
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.066294263031404e-05,
+      "loss": 1.1324,
+      "step": 6006
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.063112130456132e-05,
+      "loss": 1.4638,
+      "step": 6007
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.059930469578626e-05,
+      "loss": 1.3484,
+      "step": 6008
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.056749280780082e-05,
+      "loss": 1.0724,
+      "step": 6009
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.053568564441664e-05,
+      "loss": 1.0119,
+      "step": 6010
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0503883209444626e-05,
+      "loss": 1.1348,
+      "step": 6011
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.047208550669523e-05,
+      "loss": 1.2022,
+      "step": 6012
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.044029253997825e-05,
+      "loss": 1.2853,
+      "step": 6013
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0408504313102985e-05,
+      "loss": 1.6409,
+      "step": 6014
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0376720829878133e-05,
+      "loss": 1.6045,
+      "step": 6015
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.034494209411183e-05,
+      "loss": 1.5274,
+      "step": 6016
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.031316810961161e-05,
+      "loss": 1.4168,
+      "step": 6017
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.028139888018451e-05,
+      "loss": 1.3664,
+      "step": 6018
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.024963440963694e-05,
+      "loss": 1.4367,
+      "step": 6019
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0217874701774766e-05,
+      "loss": 1.1877,
+      "step": 6020
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.018611976040325e-05,
+      "loss": 1.3891,
+      "step": 6021
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.015436958932714e-05,
+      "loss": 1.273,
+      "step": 6022
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.012262419235053e-05,
+      "loss": 1.1203,
+      "step": 6023
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.009088357327706e-05,
+      "loss": 1.2922,
+      "step": 6024
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.005914773590966e-05,
+      "loss": 1.2009,
+      "step": 6025
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.002741668405081e-05,
+      "loss": 1.2796,
+      "step": 6026
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.99956904215023e-05,
+      "loss": 1.3818,
+      "step": 6027
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.996396895206546e-05,
+      "loss": 1.3963,
+      "step": 6028
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.993225227954101e-05,
+      "loss": 1.3848,
+      "step": 6029
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.990054040772901e-05,
+      "loss": 1.5504,
+      "step": 6030
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.986883334042909e-05,
+      "loss": 1.1644,
+      "step": 6031
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.983713108144018e-05,
+      "loss": 1.4444,
+      "step": 6032
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.98054336345607e-05,
+      "loss": 1.2399,
+      "step": 6033
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.977374100358846e-05,
+      "loss": 1.1936,
+      "step": 6034
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.974205319232074e-05,
+      "loss": 1.3939,
+      "step": 6035
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.971037020455414e-05,
+      "loss": 1.4791,
+      "step": 6036
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.967869204408487e-05,
+      "loss": 1.3804,
+      "step": 6037
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9647018714708345e-05,
+      "loss": 0.9512,
+      "step": 6038
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9615350220219556e-05,
+      "loss": 1.1913,
+      "step": 6039
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.95836865644128e-05,
+      "loss": 1.4109,
+      "step": 6040
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.955202775108193e-05,
+      "loss": 1.1686,
+      "step": 6041
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.952037378402009e-05,
+      "loss": 0.9962,
+      "step": 6042
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.948872466701994e-05,
+      "loss": 1.2045,
+      "step": 6043
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.945708040387346e-05,
+      "loss": 1.1593,
+      "step": 6044
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.942544099837217e-05,
+      "loss": 1.4622,
+      "step": 6045
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.939380645430686e-05,
+      "loss": 1.4247,
+      "step": 6046
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.936217677546788e-05,
+      "loss": 1.2753,
+      "step": 6047
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.933055196564492e-05,
+      "loss": 1.5493,
+      "step": 6048
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.929893202862711e-05,
+      "loss": 1.1866,
+      "step": 6049
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.926731696820296e-05,
+      "loss": 0.8551,
+      "step": 6050
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.923570678816046e-05,
+      "loss": 1.1149,
+      "step": 6051
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.920410149228697e-05,
+      "loss": 1.15,
+      "step": 6052
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.917250108436928e-05,
+      "loss": 1.3857,
+      "step": 6053
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.914090556819354e-05,
+      "loss": 1.2947,
+      "step": 6054
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.910931494754546e-05,
+      "loss": 1.3933,
+      "step": 6055
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9077729226209965e-05,
+      "loss": 1.4042,
+      "step": 6056
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.904614840797158e-05,
+      "loss": 1.1245,
+      "step": 6057
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.901457249661408e-05,
+      "loss": 1.2194,
+      "step": 6058
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.89830014959208e-05,
+      "loss": 1.1838,
+      "step": 6059
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8951435409674324e-05,
+      "loss": 1.1899,
+      "step": 6060
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.891987424165685e-05,
+      "loss": 1.3466,
+      "step": 6061
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.88883179956498e-05,
+      "loss": 1.4347,
+      "step": 6062
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.885676667543407e-05,
+      "loss": 1.4906,
+      "step": 6063
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.882522028479007e-05,
+      "loss": 1.2288,
+      "step": 6064
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8793678827497446e-05,
+      "loss": 1.5083,
+      "step": 6065
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.876214230733537e-05,
+      "loss": 1.1599,
+      "step": 6066
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.873061072808234e-05,
+      "loss": 1.4411,
+      "step": 6067
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8699084093516386e-05,
+      "loss": 1.204,
+      "step": 6068
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.866756240741478e-05,
+      "loss": 1.1354,
+      "step": 6069
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.863604567355436e-05,
+      "loss": 1.2933,
+      "step": 6070
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.860453389571127e-05,
+      "loss": 1.3269,
+      "step": 6071
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.85730270776611e-05,
+      "loss": 1.0355,
+      "step": 6072
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8541525223178786e-05,
+      "loss": 1.3232,
+      "step": 6073
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8510028336038804e-05,
+      "loss": 1.3457,
+      "step": 6074
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8478536420014885e-05,
+      "loss": 1.5517,
+      "step": 6075
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.844704947888028e-05,
+      "loss": 1.6138,
+      "step": 6076
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.841556751640751e-05,
+      "loss": 1.0826,
+      "step": 6077
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.838409053636869e-05,
+      "loss": 1.2735,
+      "step": 6078
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.835261854253513e-05,
+      "loss": 1.1652,
+      "step": 6079
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.832115153867778e-05,
+      "loss": 1.1381,
+      "step": 6080
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.828968952856668e-05,
+      "loss": 1.2354,
+      "step": 6081
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.825823251597157e-05,
+      "loss": 1.0172,
+      "step": 6082
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.822678050466139e-05,
+      "loss": 1.3512,
+      "step": 6083
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.819533349840465e-05,
+      "loss": 1.27,
+      "step": 6084
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.816389150096907e-05,
+      "loss": 1.3154,
+      "step": 6085
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.813245451612197e-05,
+      "loss": 1.3727,
+      "step": 6086
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.81010225476299e-05,
+      "loss": 1.3574,
+      "step": 6087
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.806959559925891e-05,
+      "loss": 1.4069,
+      "step": 6088
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.803817367477435e-05,
+      "loss": 1.2181,
+      "step": 6089
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.800675677794114e-05,
+      "loss": 1.1159,
+      "step": 6090
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.79753449125234e-05,
+      "loss": 1.398,
+      "step": 6091
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.79439380822848e-05,
+      "loss": 1.25,
+      "step": 6092
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.791253629098834e-05,
+      "loss": 1.1414,
+      "step": 6093
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.788113954239639e-05,
+      "loss": 1.4762,
+      "step": 6094
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.784974784027073e-05,
+      "loss": 1.5105,
+      "step": 6095
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.781836118837263e-05,
+      "loss": 1.2274,
+      "step": 6096
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.778697959046258e-05,
+      "loss": 1.1906,
+      "step": 6097
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.775560305030065e-05,
+      "loss": 1.5718,
+      "step": 6098
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.772423157164616e-05,
+      "loss": 1.2686,
+      "step": 6099
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.769286515825794e-05,
+      "loss": 1.4163,
+      "step": 6100
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.76615038138941e-05,
+      "loss": 1.6047,
+      "step": 6101
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.763014754231217e-05,
+      "loss": 1.4224,
+      "step": 6102
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.75987963472692e-05,
+      "loss": 1.2054,
+      "step": 6103
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.756745023252142e-05,
+      "loss": 1.3983,
+      "step": 6104
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.753610920182464e-05,
+      "loss": 1.4263,
+      "step": 6105
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7504773258933965e-05,
+      "loss": 1.3824,
+      "step": 6106
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.747344240760388e-05,
+      "loss": 1.2758,
+      "step": 6107
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7442116651588274e-05,
+      "loss": 1.7563,
+      "step": 6108
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7410795994640496e-05,
+      "loss": 1.315,
+      "step": 6109
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.737948044051318e-05,
+      "loss": 1.4154,
+      "step": 6110
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7348169992958424e-05,
+      "loss": 1.3858,
+      "step": 6111
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7316864655727644e-05,
+      "loss": 1.3402,
+      "step": 6112
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.728556443257176e-05,
+      "loss": 1.2659,
+      "step": 6113
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.725426932724095e-05,
+      "loss": 1.1799,
+      "step": 6114
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.722297934348485e-05,
+      "loss": 1.4591,
+      "step": 6115
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7191694485052415e-05,
+      "loss": 1.1747,
+      "step": 6116
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.716041475569213e-05,
+      "loss": 1.372,
+      "step": 6117
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.7129140159151674e-05,
+      "loss": 1.248,
+      "step": 6118
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.709787069917829e-05,
+      "loss": 1.1425,
+      "step": 6119
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.70666063795185e-05,
+      "loss": 1.2461,
+      "step": 6120
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.703534720391823e-05,
+      "loss": 1.5188,
+      "step": 6121
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.700409317612276e-05,
+      "loss": 1.4189,
+      "step": 6122
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.697284429987685e-05,
+      "loss": 1.2739,
+      "step": 6123
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6941600578924506e-05,
+      "loss": 1.2256,
+      "step": 6124
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6910362017009275e-05,
+      "loss": 1.1126,
+      "step": 6125
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.687912861787392e-05,
+      "loss": 1.5479,
+      "step": 6126
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6847900385260754e-05,
+      "loss": 1.2652,
+      "step": 6127
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.681667732291134e-05,
+      "loss": 1.3302,
+      "step": 6128
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.678545943456666e-05,
+      "loss": 1.2346,
+      "step": 6129
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6754246723967055e-05,
+      "loss": 1.4932,
+      "step": 6130
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6723039194852325e-05,
+      "loss": 1.4334,
+      "step": 6131
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6691836850961544e-05,
+      "loss": 1.4469,
+      "step": 6132
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.666063969603328e-05,
+      "loss": 1.1181,
+      "step": 6133
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.66294477338054e-05,
+      "loss": 1.1607,
+      "step": 6134
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.659826096801509e-05,
+      "loss": 1.2149,
+      "step": 6135
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6567079402399094e-05,
+      "loss": 1.3504,
+      "step": 6136
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.653590304069335e-05,
+      "loss": 1.1986,
+      "step": 6137
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.650473188663331e-05,
+      "loss": 1.235,
+      "step": 6138
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6473565943953674e-05,
+      "loss": 1.5466,
+      "step": 6139
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6442405216388716e-05,
+      "loss": 1.4683,
+      "step": 6140
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6411249707671775e-05,
+      "loss": 1.2378,
+      "step": 6141
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.638009942153588e-05,
+      "loss": 1.4579,
+      "step": 6142
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6348954361713216e-05,
+      "loss": 1.1518,
+      "step": 6143
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6317814531935495e-05,
+      "loss": 1.2915,
+      "step": 6144
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.628667993593366e-05,
+      "loss": 1.2498,
+      "step": 6145
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.625555057743816e-05,
+      "loss": 1.5659,
+      "step": 6146
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.622442646017876e-05,
+      "loss": 1.2612,
+      "step": 6147
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.619330758788455e-05,
+      "loss": 1.4671,
+      "step": 6148
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6162193964284015e-05,
+      "loss": 1.4002,
+      "step": 6149
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6131085593105115e-05,
+      "loss": 1.5198,
+      "step": 6150
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.609998247807498e-05,
+      "loss": 1.1024,
+      "step": 6151
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6068884622920346e-05,
+      "loss": 1.2646,
+      "step": 6152
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.603779203136714e-05,
+      "loss": 1.3295,
+      "step": 6153
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.600670470714073e-05,
+      "loss": 1.1129,
+      "step": 6154
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.597562265396579e-05,
+      "loss": 1.4116,
+      "step": 6155
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.59445458755665e-05,
+      "loss": 1.2456,
+      "step": 6156
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.591347437566622e-05,
+      "loss": 1.6075,
+      "step": 6157
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.5882408157987875e-05,
+      "loss": 1.3277,
+      "step": 6158
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.585134722625358e-05,
+      "loss": 1.4865,
+      "step": 6159
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.582029158418498e-05,
+      "loss": 1.136,
+      "step": 6160
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.578924123550295e-05,
+      "loss": 1.7394,
+      "step": 6161
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.5758196183927804e-05,
+      "loss": 1.2114,
+      "step": 6162
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.572715643317914e-05,
+      "loss": 1.3365,
+      "step": 6163
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.5696121986976066e-05,
+      "loss": 1.3814,
+      "step": 6164
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.5665092849036914e-05,
+      "loss": 1.2216,
+      "step": 6165
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.5634069023079485e-05,
+      "loss": 1.3275,
+      "step": 6166
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.5603050512820864e-05,
+      "loss": 1.2962,
+      "step": 6167
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.5572037321977556e-05,
+      "loss": 1.222,
+      "step": 6168
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.5541029454265316e-05,
+      "loss": 1.1923,
+      "step": 6169
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.551002691339947e-05,
+      "loss": 1.4274,
+      "step": 6170
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.5479029703094485e-05,
+      "loss": 1.3931,
+      "step": 6171
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.5448037827064325e-05,
+      "loss": 1.1391,
+      "step": 6172
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.541705128902233e-05,
+      "loss": 1.6188,
+      "step": 6173
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.5386070092681084e-05,
+      "loss": 1.628,
+      "step": 6174
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.535509424175263e-05,
+      "loss": 1.372,
+      "step": 6175
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.5324123739948275e-05,
+      "loss": 1.0977,
+      "step": 6176
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.529315859097883e-05,
+      "loss": 1.0783,
+      "step": 6177
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.52621987985543e-05,
+      "loss": 1.2768,
+      "step": 6178
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.523124436638422e-05,
+      "loss": 1.2338,
+      "step": 6179
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.520029529817733e-05,
+      "loss": 1.2239,
+      "step": 6180
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.51693515976418e-05,
+      "loss": 1.308,
+      "step": 6181
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.51384132684851e-05,
+      "loss": 1.1596,
+      "step": 6182
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.5107480314414204e-05,
+      "loss": 1.2789,
+      "step": 6183
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.507655273913524e-05,
+      "loss": 1.4704,
+      "step": 6184
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.504563054635388e-05,
+      "loss": 1.3035,
+      "step": 6185
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.501471373977497e-05,
+      "loss": 1.0662,
+      "step": 6186
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.498380232310289e-05,
+      "loss": 1.2366,
+      "step": 6187
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.495289630004126e-05,
+      "loss": 1.5855,
+      "step": 6188
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.492199567429307e-05,
+      "loss": 1.4588,
+      "step": 6189
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4891100449560636e-05,
+      "loss": 1.1269,
+      "step": 6190
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4860210629545747e-05,
+      "loss": 1.3516,
+      "step": 6191
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.482932621794938e-05,
+      "loss": 1.382,
+      "step": 6192
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4798447218472046e-05,
+      "loss": 1.1686,
+      "step": 6193
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4767573634813443e-05,
+      "loss": 1.2952,
+      "step": 6194
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.473670547067269e-05,
+      "loss": 1.6055,
+      "step": 6195
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4705842729748214e-05,
+      "loss": 1.5492,
+      "step": 6196
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4674985415737926e-05,
+      "loss": 0.9601,
+      "step": 6197
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.46441335323389e-05,
+      "loss": 1.0732,
+      "step": 6198
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4613287083247725e-05,
+      "loss": 1.6097,
+      "step": 6199
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4582446072160235e-05,
+      "loss": 0.9382,
+      "step": 6200
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4551610502771634e-05,
+      "loss": 1.1163,
+      "step": 6201
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.452078037877645e-05,
+      "loss": 1.4966,
+      "step": 6202
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.448995570386866e-05,
+      "loss": 1.5286,
+      "step": 6203
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.445913648174146e-05,
+      "loss": 1.4479,
+      "step": 6204
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.44283227160875e-05,
+      "loss": 1.4062,
+      "step": 6205
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.439751441059867e-05,
+      "loss": 1.4312,
+      "step": 6206
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4366711568966335e-05,
+      "loss": 1.4698,
+      "step": 6207
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.433591419488111e-05,
+      "loss": 0.9853,
+      "step": 6208
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.430512229203291e-05,
+      "loss": 1.2998,
+      "step": 6209
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.427433586411117e-05,
+      "loss": 1.1264,
+      "step": 6210
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4243554914804464e-05,
+      "loss": 1.2905,
+      "step": 6211
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.42127794478009e-05,
+      "loss": 1.1953,
+      "step": 6212
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.41820094667878e-05,
+      "loss": 1.2961,
+      "step": 6213
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.415124497545184e-05,
+      "loss": 0.985,
+      "step": 6214
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.4120485977479054e-05,
+      "loss": 1.3818,
+      "step": 6215
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.408973247655489e-05,
+      "loss": 1.5729,
+      "step": 6216
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.4058984476364006e-05,
+      "loss": 1.1074,
+      "step": 6217
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.402824198059053e-05,
+      "loss": 1.2192,
+      "step": 6218
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.399750499291782e-05,
+      "loss": 1.2332,
+      "step": 6219
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.396677351702869e-05,
+      "loss": 1.1875,
+      "step": 6220
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.393604755660517e-05,
+      "loss": 1.4164,
+      "step": 6221
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.390532711532873e-05,
+      "loss": 1.2951,
+      "step": 6222
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3874612196880056e-05,
+      "loss": 0.8082,
+      "step": 6223
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.384390280493935e-05,
+      "loss": 1.3079,
+      "step": 6224
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.381319894318597e-05,
+      "loss": 1.136,
+      "step": 6225
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.378250061529879e-05,
+      "loss": 1.2311,
+      "step": 6226
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3751807824955855e-05,
+      "loss": 1.1839,
+      "step": 6227
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.372112057583465e-05,
+      "loss": 1.3216,
+      "step": 6228
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3690438871611906e-05,
+      "loss": 1.4321,
+      "step": 6229
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.365976271596382e-05,
+      "loss": 1.0843,
+      "step": 6230
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3629092112565816e-05,
+      "loss": 1.3653,
+      "step": 6231
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.359842706509271e-05,
+      "loss": 1.2748,
+      "step": 6232
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.35677675772186e-05,
+      "loss": 1.1658,
+      "step": 6233
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.353711365261699e-05,
+      "loss": 1.4191,
+      "step": 6234
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.350646529496066e-05,
+      "loss": 1.2471,
+      "step": 6235
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3475822507921735e-05,
+      "loss": 1.1329,
+      "step": 6236
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.344518529517163e-05,
+      "loss": 1.1035,
+      "step": 6237
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.341455366038123e-05,
+      "loss": 1.2645,
+      "step": 6238
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3383927607220575e-05,
+      "loss": 1.6363,
+      "step": 6239
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3353307139359196e-05,
+      "loss": 1.4945,
+      "step": 6240
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3322692260465845e-05,
+      "loss": 1.0884,
+      "step": 6241
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3292082974208644e-05,
+      "loss": 1.6361,
+      "step": 6242
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.326147928425501e-05,
+      "loss": 1.3928,
+      "step": 6243
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.323088119427175e-05,
+      "loss": 1.3762,
+      "step": 6244
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3200288707925e-05,
+      "loss": 1.1792,
+      "step": 6245
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3169701828880126e-05,
+      "loss": 1.2245,
+      "step": 6246
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.313912056080197e-05,
+      "loss": 0.9133,
+      "step": 6247
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3108544907354583e-05,
+      "loss": 1.481,
+      "step": 6248
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3077974872201384e-05,
+      "loss": 1.2798,
+      "step": 6249
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3047410459005084e-05,
+      "loss": 1.2337,
+      "step": 6250
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.301685167142783e-05,
+      "loss": 1.5355,
+      "step": 6251
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.2986298513130925e-05,
+      "loss": 1.2711,
+      "step": 6252
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.29557509877752e-05,
+      "loss": 0.8907,
+      "step": 6253
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.2925209099020636e-05,
+      "loss": 1.4097,
+      "step": 6254
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.289467285052663e-05,
+      "loss": 1.6936,
+      "step": 6255
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.286414224595182e-05,
+      "loss": 1.0192,
+      "step": 6256
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.283361728895432e-05,
+      "loss": 1.4571,
+      "step": 6257
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.280309798319139e-05,
+      "loss": 1.1542,
+      "step": 6258
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.277258433231975e-05,
+      "loss": 1.5036,
+      "step": 6259
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.274207633999539e-05,
+      "loss": 1.2189,
+      "step": 6260
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.271157400987362e-05,
+      "loss": 1.0975,
+      "step": 6261
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.268107734560901e-05,
+      "loss": 1.3269,
+      "step": 6262
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.2650586350855604e-05,
+      "loss": 1.2234,
+      "step": 6263
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.26201010292666e-05,
+      "loss": 1.2439,
+      "step": 6264
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.258962138449466e-05,
+      "loss": 1.5912,
+      "step": 6265
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.255914742019164e-05,
+      "loss": 0.9577,
+      "step": 6266
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.252867914000883e-05,
+      "loss": 1.4526,
+      "step": 6267
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.249821654759677e-05,
+      "loss": 1.227,
+      "step": 6268
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.246775964660531e-05,
+      "loss": 1.1833,
+      "step": 6269
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.243730844068362e-05,
+      "loss": 0.9728,
+      "step": 6270
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.240686293348026e-05,
+      "loss": 1.1724,
+      "step": 6271
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.237642312864299e-05,
+      "loss": 1.4477,
+      "step": 6272
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.2345989029819056e-05,
+      "loss": 1.6329,
+      "step": 6273
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.2315560640654835e-05,
+      "loss": 1.462,
+      "step": 6274
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.228513796479613e-05,
+      "loss": 1.2479,
+      "step": 6275
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.225472100588796e-05,
+      "loss": 1.0634,
+      "step": 6276
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.222430976757484e-05,
+      "loss": 1.4056,
+      "step": 6277
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.2193904253500406e-05,
+      "loss": 1.5396,
+      "step": 6278
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.216350446730771e-05,
+      "loss": 1.0879,
+      "step": 6279
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.213311041263915e-05,
+      "loss": 1.2292,
+      "step": 6280
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.210272209313636e-05,
+      "loss": 1.465,
+      "step": 6281
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.207233951244028e-05,
+      "loss": 1.3348,
+      "step": 6282
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.204196267419119e-05,
+      "loss": 1.1892,
+      "step": 6283
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.201159158202876e-05,
+      "loss": 1.4686,
+      "step": 6284
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.198122623959181e-05,
+      "loss": 1.187,
+      "step": 6285
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1950866650518625e-05,
+      "loss": 1.2693,
+      "step": 6286
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.192051281844672e-05,
+      "loss": 1.5324,
+      "step": 6287
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.189016474701294e-05,
+      "loss": 1.2962,
+      "step": 6288
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.185982243985337e-05,
+      "loss": 1.2906,
+      "step": 6289
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.182948590060357e-05,
+      "loss": 1.2371,
+      "step": 6290
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1799155132898216e-05,
+      "loss": 1.1724,
+      "step": 6291
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.176883014037148e-05,
+      "loss": 1.114,
+      "step": 6292
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.173851092665666e-05,
+      "loss": 1.3003,
+      "step": 6293
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.170819749538652e-05,
+      "loss": 1.2723,
+      "step": 6294
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.167788985019304e-05,
+      "loss": 1.0218,
+      "step": 6295
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1647587994707505e-05,
+      "loss": 1.2816,
+      "step": 6296
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.161729193256051e-05,
+      "loss": 1.4532,
+      "step": 6297
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.158700166738203e-05,
+      "loss": 1.4118,
+      "step": 6298
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1556717202801216e-05,
+      "loss": 1.4056,
+      "step": 6299
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.152643854244671e-05,
+      "loss": 1.4358,
+      "step": 6300
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.149616568994626e-05,
+      "loss": 0.9925,
+      "step": 6301
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.146589864892704e-05,
+      "loss": 1.221,
+      "step": 6302
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1435637423015424e-05,
+      "loss": 1.1618,
+      "step": 6303
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1405382015837244e-05,
+      "loss": 1.3386,
+      "step": 6304
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.137513243101749e-05,
+      "loss": 1.5647,
+      "step": 6305
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.134488867218057e-05,
+      "loss": 1.2605,
+      "step": 6306
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.131465074295006e-05,
+      "loss": 1.5339,
+      "step": 6307
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.128441864694904e-05,
+      "loss": 1.2751,
+      "step": 6308
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.125419238779961e-05,
+      "loss": 1.1456,
+      "step": 6309
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.122397196912343e-05,
+      "loss": 1.1363,
+      "step": 6310
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.119375739454127e-05,
+      "loss": 1.2637,
+      "step": 6311
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.1163548667673386e-05,
+      "loss": 1.2907,
+      "step": 6312
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.113334579213915e-05,
+      "loss": 1.1248,
+      "step": 6313
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.11031487715574e-05,
+      "loss": 0.9602,
+      "step": 6314
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.107295760954613e-05,
+      "loss": 1.5639,
+      "step": 6315
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.1042772309722675e-05,
+      "loss": 1.5635,
+      "step": 6316
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.101259287570374e-05,
+      "loss": 1.3469,
+      "step": 6317
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0982419311105186e-05,
+      "loss": 1.2258,
+      "step": 6318
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.095225161954237e-05,
+      "loss": 1.5813,
+      "step": 6319
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.092208980462976e-05,
+      "loss": 1.1739,
+      "step": 6320
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.089193386998119e-05,
+      "loss": 1.314,
+      "step": 6321
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.086178381920977e-05,
+      "loss": 1.2333,
+      "step": 6322
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0831639655927986e-05,
+      "loss": 1.1958,
+      "step": 6323
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0801501383747505e-05,
+      "loss": 1.4389,
+      "step": 6324
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.077136900627938e-05,
+      "loss": 1.1727,
+      "step": 6325
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.074124252713387e-05,
+      "loss": 0.989,
+      "step": 6326
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.071112194992064e-05,
+      "loss": 1.4282,
+      "step": 6327
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.068100727824855e-05,
+      "loss": 1.1724,
+      "step": 6328
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0650898515725776e-05,
+      "loss": 1.1694,
+      "step": 6329
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0620795665959774e-05,
+      "loss": 1.7327,
+      "step": 6330
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0590698732557384e-05,
+      "loss": 1.8012,
+      "step": 6331
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0560607719124586e-05,
+      "loss": 1.3056,
+      "step": 6332
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.053052262926681e-05,
+      "loss": 1.5798,
+      "step": 6333
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.050044346658864e-05,
+      "loss": 1.2001,
+      "step": 6334
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0470370234694054e-05,
+      "loss": 1.5619,
+      "step": 6335
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.044030293718618e-05,
+      "loss": 1.3407,
+      "step": 6336
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.041024157766765e-05,
+      "loss": 1.3941,
+      "step": 6337
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.038018615974016e-05,
+      "loss": 1.569,
+      "step": 6338
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.035013668700488e-05,
+      "loss": 1.1799,
+      "step": 6339
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0320093163062096e-05,
+      "loss": 1.3477,
+      "step": 6340
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.029005559151157e-05,
+      "loss": 1.1735,
+      "step": 6341
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0260023975952195e-05,
+      "loss": 1.3982,
+      "step": 6342
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0229998319982206e-05,
+      "loss": 1.3365,
+      "step": 6343
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.01999786271991e-05,
+      "loss": 1.2916,
+      "step": 6344
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.016996490119974e-05,
+      "loss": 1.226,
+      "step": 6345
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.013995714558015e-05,
+      "loss": 1.2225,
+      "step": 6346
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0109955363935766e-05,
+      "loss": 1.1582,
+      "step": 6347
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.007995955986124e-05,
+      "loss": 0.8102,
+      "step": 6348
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.00499697369505e-05,
+      "loss": 1.2409,
+      "step": 6349
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.001998589879673e-05,
+      "loss": 1.444,
+      "step": 6350
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.9990008048992476e-05,
+      "loss": 1.0545,
+      "step": 6351
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.996003619112959e-05,
+      "loss": 1.1853,
+      "step": 6352
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.993007032879905e-05,
+      "loss": 1.2776,
+      "step": 6353
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9900110465591284e-05,
+      "loss": 1.2545,
+      "step": 6354
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.987015660509592e-05,
+      "loss": 1.1923,
+      "step": 6355
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.984020875090184e-05,
+      "loss": 1.2985,
+      "step": 6356
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.981026690659724e-05,
+      "loss": 1.4335,
+      "step": 6357
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.978033107576965e-05,
+      "loss": 1.2567,
+      "step": 6358
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.975040126200575e-05,
+      "loss": 1.4694,
+      "step": 6359
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.972047746889168e-05,
+      "loss": 1.2541,
+      "step": 6360
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.969055970001268e-05,
+      "loss": 1.316,
+      "step": 6361
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.966064795895339e-05,
+      "loss": 1.3671,
+      "step": 6362
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.96307422492976e-05,
+      "loss": 1.0216,
+      "step": 6363
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.960084257462856e-05,
+      "loss": 1.3096,
+      "step": 6364
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.957094893852861e-05,
+      "loss": 1.0981,
+      "step": 6365
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9541061344579534e-05,
+      "loss": 1.3348,
+      "step": 6366
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.951117979636223e-05,
+      "loss": 1.5125,
+      "step": 6367
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.948130429745707e-05,
+      "loss": 1.427,
+      "step": 6368
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9451434851443426e-05,
+      "loss": 1.1988,
+      "step": 6369
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.942157146190022e-05,
+      "loss": 1.0929,
+      "step": 6370
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9391714132405454e-05,
+      "loss": 0.9852,
+      "step": 6371
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9361862866536554e-05,
+      "loss": 1.1273,
+      "step": 6372
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.933201766787009e-05,
+      "loss": 1.06,
+      "step": 6373
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9302178539982e-05,
+      "loss": 1.2844,
+      "step": 6374
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9272345486447446e-05,
+      "loss": 0.8842,
+      "step": 6375
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9242518510840865e-05,
+      "loss": 1.3892,
+      "step": 6376
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.921269761673594e-05,
+      "loss": 1.4568,
+      "step": 6377
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.918288280770571e-05,
+      "loss": 1.4399,
+      "step": 6378
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.915307408732238e-05,
+      "loss": 1.3638,
+      "step": 6379
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.912327145915755e-05,
+      "loss": 1.5383,
+      "step": 6380
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.909347492678198e-05,
+      "loss": 1.199,
+      "step": 6381
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.906368449376574e-05,
+      "loss": 1.416,
+      "step": 6382
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.903390016367812e-05,
+      "loss": 1.1619,
+      "step": 6383
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.90041219400878e-05,
+      "loss": 1.3401,
+      "step": 6384
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.897434982656259e-05,
+      "loss": 1.2532,
+      "step": 6385
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.894458382666971e-05,
+      "loss": 1.103,
+      "step": 6386
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8914823943975486e-05,
+      "loss": 1.5634,
+      "step": 6387
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8885070182045665e-05,
+      "loss": 1.1057,
+      "step": 6388
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.885532254444517e-05,
+      "loss": 1.2027,
+      "step": 6389
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8825581034738167e-05,
+      "loss": 1.1807,
+      "step": 6390
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.879584565648819e-05,
+      "loss": 1.6424,
+      "step": 6391
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.876611641325792e-05,
+      "loss": 1.4992,
+      "step": 6392
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.873639330860945e-05,
+      "loss": 1.225,
+      "step": 6393
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8706676346103985e-05,
+      "loss": 1.3259,
+      "step": 6394
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.867696552930208e-05,
+      "loss": 0.8701,
+      "step": 6395
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8647260861763475e-05,
+      "loss": 1.4108,
+      "step": 6396
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.861756234704734e-05,
+      "loss": 1.0365,
+      "step": 6397
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8587869988711884e-05,
+      "loss": 1.2676,
+      "step": 6398
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.855818379031481e-05,
+      "loss": 1.2146,
+      "step": 6399
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.8528503755412845e-05,
+      "loss": 1.1792,
+      "step": 6400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.849882988756222e-05,
+      "loss": 1.5635,
+      "step": 6401
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.846916219031823e-05,
+      "loss": 1.1979,
+      "step": 6402
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.8439500667235536e-05,
+      "loss": 1.1889,
+      "step": 6403
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.8409845321867975e-05,
+      "loss": 1.0226,
+      "step": 6404
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.8380196157768774e-05,
+      "loss": 1.0974,
+      "step": 6405
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.8350553178490275e-05,
+      "loss": 1.5126,
+      "step": 6406
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.8320916387584235e-05,
+      "loss": 1.2205,
+      "step": 6407
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.829128578860151e-05,
+      "loss": 0.9942,
+      "step": 6408
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.826166138509233e-05,
+      "loss": 0.8737,
+      "step": 6409
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.8232043180606065e-05,
+      "loss": 1.5032,
+      "step": 6410
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.820243117869152e-05,
+      "loss": 1.3688,
+      "step": 6411
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.817282538289656e-05,
+      "loss": 1.1369,
+      "step": 6412
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.8143225796768486e-05,
+      "loss": 1.1124,
+      "step": 6413
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.811363242385368e-05,
+      "loss": 1.3677,
+      "step": 6414
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.8084045267698e-05,
+      "loss": 1.6045,
+      "step": 6415
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.805446433184625e-05,
+      "loss": 1.3283,
+      "step": 6416
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.802488961984282e-05,
+      "loss": 1.1773,
+      "step": 6417
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.799532113523109e-05,
+      "loss": 1.3948,
+      "step": 6418
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.796575888155388e-05,
+      "loss": 1.5198,
+      "step": 6419
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.793620286235314e-05,
+      "loss": 1.5017,
+      "step": 6420
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.790665308117017e-05,
+      "loss": 1.51,
+      "step": 6421
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7877109541545454e-05,
+      "loss": 1.2916,
+      "step": 6422
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.78475722470187e-05,
+      "loss": 1.5884,
+      "step": 6423
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.781804120112899e-05,
+      "loss": 1.4643,
+      "step": 6424
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.778851640741452e-05,
+      "loss": 0.8766,
+      "step": 6425
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.775899786941286e-05,
+      "loss": 1.1579,
+      "step": 6426
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.77294855906607e-05,
+      "loss": 1.4129,
+      "step": 6427
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7699979574694165e-05,
+      "loss": 1.277,
+      "step": 6428
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.767047982504836e-05,
+      "loss": 0.8879,
+      "step": 6429
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.764098634525792e-05,
+      "loss": 1.1946,
+      "step": 6430
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.761149913885651e-05,
+      "loss": 1.5011,
+      "step": 6431
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.758201820937722e-05,
+      "loss": 1.3961,
+      "step": 6432
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.755254356035223e-05,
+      "loss": 1.1774,
+      "step": 6433
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.75230751953131e-05,
+      "loss": 1.482,
+      "step": 6434
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.749361311779058e-05,
+      "loss": 1.2747,
+      "step": 6435
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.746415733131463e-05,
+      "loss": 1.3686,
+      "step": 6436
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.743470783941446e-05,
+      "loss": 1.418,
+      "step": 6437
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.740526464561864e-05,
+      "loss": 1.3467,
+      "step": 6438
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.737582775345483e-05,
+      "loss": 0.9159,
+      "step": 6439
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7346397166450075e-05,
+      "loss": 1.0249,
+      "step": 6440
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7316972888130564e-05,
+      "loss": 1.4482,
+      "step": 6441
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.728755492202176e-05,
+      "loss": 1.1501,
+      "step": 6442
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.725814327164835e-05,
+      "loss": 1.1005,
+      "step": 6443
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.722873794053434e-05,
+      "loss": 1.5178,
+      "step": 6444
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.719933893220286e-05,
+      "loss": 1.3569,
+      "step": 6445
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7169946250176436e-05,
+      "loss": 1.3553,
+      "step": 6446
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.714055989797664e-05,
+      "loss": 1.4906,
+      "step": 6447
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7111179879124525e-05,
+      "loss": 1.3397,
+      "step": 6448
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7081806197140166e-05,
+      "loss": 1.6349,
+      "step": 6449
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.705243885554299e-05,
+      "loss": 1.1725,
+      "step": 6450
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7023077857851595e-05,
+      "loss": 1.3392,
+      "step": 6451
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.699372320758395e-05,
+      "loss": 1.5209,
+      "step": 6452
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.69643749082571e-05,
+      "loss": 1.0885,
+      "step": 6453
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.693503296338748e-05,
+      "loss": 1.1639,
+      "step": 6454
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.690569737649067e-05,
+      "loss": 1.5122,
+      "step": 6455
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6876368151081475e-05,
+      "loss": 1.4209,
+      "step": 6456
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6847045290673976e-05,
+      "loss": 1.3974,
+      "step": 6457
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.681772879878153e-05,
+      "loss": 1.2899,
+      "step": 6458
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6788418678916636e-05,
+      "loss": 1.411,
+      "step": 6459
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.675911493459111e-05,
+      "loss": 1.4104,
+      "step": 6460
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.672981756931601e-05,
+      "loss": 1.1711,
+      "step": 6461
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.670052658660158e-05,
+      "loss": 1.2623,
+      "step": 6462
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.667124198995728e-05,
+      "loss": 1.3488,
+      "step": 6463
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.664196378289184e-05,
+      "loss": 1.0807,
+      "step": 6464
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.661269196891328e-05,
+      "loss": 1.4336,
+      "step": 6465
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.658342655152873e-05,
+      "loss": 1.137,
+      "step": 6466
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.65541675342447e-05,
+      "loss": 1.3316,
+      "step": 6467
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6524914920566795e-05,
+      "loss": 1.0823,
+      "step": 6468
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6495668713999953e-05,
+      "loss": 0.9552,
+      "step": 6469
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.646642891804823e-05,
+      "loss": 1.3945,
+      "step": 6470
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6437195536215084e-05,
+      "loss": 1.278,
+      "step": 6471
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6407968572003036e-05,
+      "loss": 1.0688,
+      "step": 6472
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.637874802891398e-05,
+      "loss": 1.2435,
+      "step": 6473
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.634953391044889e-05,
+      "loss": 1.6244,
+      "step": 6474
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.632032622010818e-05,
+      "loss": 1.4801,
+      "step": 6475
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6291124961391205e-05,
+      "loss": 1.4318,
+      "step": 6476
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6261930137796825e-05,
+      "loss": 1.2414,
+      "step": 6477
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.623274175282294e-05,
+      "loss": 1.0548,
+      "step": 6478
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.620355980996683e-05,
+      "loss": 1.4313,
+      "step": 6479
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.617438431272485e-05,
+      "loss": 1.254,
+      "step": 6480
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.614521526459273e-05,
+      "loss": 1.2572,
+      "step": 6481
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.611605266906533e-05,
+      "loss": 1.1528,
+      "step": 6482
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.608689652963675e-05,
+      "loss": 1.4235,
+      "step": 6483
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.605774684980029e-05,
+      "loss": 1.3647,
+      "step": 6484
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6028603633048596e-05,
+      "loss": 1.3194,
+      "step": 6485
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.59994668828734e-05,
+      "loss": 1.2268,
+      "step": 6486
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.597033660276576e-05,
+      "loss": 1.2004,
+      "step": 6487
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5941212796215916e-05,
+      "loss": 1.1081,
+      "step": 6488
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.59120954667133e-05,
+      "loss": 1.3701,
+      "step": 6489
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5882984617746594e-05,
+      "loss": 1.0173,
+      "step": 6490
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.585388025280376e-05,
+      "loss": 1.3189,
+      "step": 6491
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.582478237537188e-05,
+      "loss": 1.1063,
+      "step": 6492
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5795690988937365e-05,
+      "loss": 1.6527,
+      "step": 6493
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.576660609698574e-05,
+      "loss": 1.116,
+      "step": 6494
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.573752770300188e-05,
+      "loss": 1.1653,
+      "step": 6495
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.570845581046976e-05,
+      "loss": 0.9613,
+      "step": 6496
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5679390422872614e-05,
+      "loss": 1.4278,
+      "step": 6497
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5650331543692956e-05,
+      "loss": 1.2681,
+      "step": 6498
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.562127917641241e-05,
+      "loss": 1.1143,
+      "step": 6499
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5592233324511946e-05,
+      "loss": 1.0933,
+      "step": 6500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.556319399147167e-05,
+      "loss": 1.4216,
+      "step": 6501
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.553416118077091e-05,
+      "loss": 1.1105,
+      "step": 6502
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.55051348958882e-05,
+      "loss": 1.6536,
+      "step": 6503
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.547611514030139e-05,
+      "loss": 0.9386,
+      "step": 6504
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.544710191748741e-05,
+      "loss": 1.4088,
+      "step": 6505
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.541809523092255e-05,
+      "loss": 1.2515,
+      "step": 6506
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5389095084082164e-05,
+      "loss": 1.0533,
+      "step": 6507
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.536010148044097e-05,
+      "loss": 1.4452,
+      "step": 6508
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.533111442347282e-05,
+      "loss": 1.1311,
+      "step": 6509
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5302133916650766e-05,
+      "loss": 1.2245,
+      "step": 6510
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.527315996344709e-05,
+      "loss": 1.3324,
+      "step": 6511
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.524419256733335e-05,
+      "loss": 1.3193,
+      "step": 6512
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.521523173178023e-05,
+      "loss": 1.5061,
+      "step": 6513
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.518627746025772e-05,
+      "loss": 1.3292,
+      "step": 6514
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5157329756234933e-05,
+      "loss": 1.5191,
+      "step": 6515
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.512838862318026e-05,
+      "loss": 1.1395,
+      "step": 6516
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.509945406456122e-05,
+      "loss": 1.3244,
+      "step": 6517
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.507052608384468e-05,
+      "loss": 1.6329,
+      "step": 6518
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.504160468449658e-05,
+      "loss": 1.4746,
+      "step": 6519
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5012689869982194e-05,
+      "loss": 1.0323,
+      "step": 6520
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.498378164376588e-05,
+      "loss": 1.2853,
+      "step": 6521
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.495488000931138e-05,
+      "loss": 1.357,
+      "step": 6522
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.49259849700814e-05,
+      "loss": 1.2142,
+      "step": 6523
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.489709652953811e-05,
+      "loss": 1.4478,
+      "step": 6524
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4868214691142696e-05,
+      "loss": 1.1296,
+      "step": 6525
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.48393394583557e-05,
+      "loss": 1.328,
+      "step": 6526
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4810470834636734e-05,
+      "loss": 0.8046,
+      "step": 6527
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.478160882344477e-05,
+      "loss": 1.8158,
+      "step": 6528
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.475275342823785e-05,
+      "loss": 1.3223,
+      "step": 6529
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4723904652473306e-05,
+      "loss": 1.3798,
+      "step": 6530
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4695062499607586e-05,
+      "loss": 1.2565,
+      "step": 6531
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.466622697309647e-05,
+      "loss": 1.4669,
+      "step": 6532
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.463739807639492e-05,
+      "loss": 1.4148,
+      "step": 6533
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.460857581295698e-05,
+      "loss": 1.4249,
+      "step": 6534
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4579760186236084e-05,
+      "loss": 1.4717,
+      "step": 6535
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.455095119968466e-05,
+      "loss": 1.3241,
+      "step": 6536
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4522148856754544e-05,
+      "loss": 1.2629,
+      "step": 6537
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4493353160896613e-05,
+      "loss": 1.3265,
+      "step": 6538
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4464564115561094e-05,
+      "loss": 1.105,
+      "step": 6539
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.4435781724197266e-05,
+      "loss": 1.5085,
+      "step": 6540
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.440700599025377e-05,
+      "loss": 1.4975,
+      "step": 6541
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.437823691717833e-05,
+      "loss": 1.2345,
+      "step": 6542
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.43494745084179e-05,
+      "loss": 1.1419,
+      "step": 6543
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.432071876741861e-05,
+      "loss": 1.2186,
+      "step": 6544
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.42919696976259e-05,
+      "loss": 1.6209,
+      "step": 6545
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.4263227302484275e-05,
+      "loss": 1.3799,
+      "step": 6546
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.4234491585437555e-05,
+      "loss": 1.1016,
+      "step": 6547
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.4205762549928686e-05,
+      "loss": 1.4162,
+      "step": 6548
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.417704019939983e-05,
+      "loss": 1.8176,
+      "step": 6549
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.414832453729231e-05,
+      "loss": 1.1794,
+      "step": 6550
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.411961556704677e-05,
+      "loss": 1.1861,
+      "step": 6551
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.409091329210289e-05,
+      "loss": 1.4199,
+      "step": 6552
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.406221771589971e-05,
+      "loss": 1.3772,
+      "step": 6553
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.403352884187532e-05,
+      "loss": 1.4894,
+      "step": 6554
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.400484667346712e-05,
+      "loss": 1.385,
+      "step": 6555
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.397617121411166e-05,
+      "loss": 1.0063,
+      "step": 6556
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.394750246724466e-05,
+      "loss": 1.6776,
+      "step": 6557
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.391884043630105e-05,
+      "loss": 0.8087,
+      "step": 6558
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.389018512471501e-05,
+      "loss": 0.9096,
+      "step": 6559
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3861536535919825e-05,
+      "loss": 0.9737,
+      "step": 6560
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.383289467334808e-05,
+      "loss": 1.0872,
+      "step": 6561
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.380425954043148e-05,
+      "loss": 1.3109,
+      "step": 6562
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3775631140600916e-05,
+      "loss": 1.1837,
+      "step": 6563
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.374700947728647e-05,
+      "loss": 0.9662,
+      "step": 6564
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.371839455391753e-05,
+      "loss": 1.1675,
+      "step": 6565
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.368978637392249e-05,
+      "loss": 1.3424,
+      "step": 6566
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3661184940729093e-05,
+      "loss": 0.8392,
+      "step": 6567
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.363259025776425e-05,
+      "loss": 1.295,
+      "step": 6568
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3604002328453976e-05,
+      "loss": 1.2967,
+      "step": 6569
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.357542115622355e-05,
+      "loss": 1.3063,
+      "step": 6570
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.354684674449738e-05,
+      "loss": 1.1586,
+      "step": 6571
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3518279096699174e-05,
+      "loss": 1.2914,
+      "step": 6572
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.348971821625169e-05,
+      "loss": 1.1675,
+      "step": 6573
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.346116410657704e-05,
+      "loss": 1.2674,
+      "step": 6574
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3432616771096354e-05,
+      "loss": 1.4541,
+      "step": 6575
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.340407621323006e-05,
+      "loss": 1.3669,
+      "step": 6576
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3375542436397697e-05,
+      "loss": 1.4116,
+      "step": 6577
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3347015444018104e-05,
+      "loss": 1.3978,
+      "step": 6578
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.331849523950917e-05,
+      "loss": 1.3786,
+      "step": 6579
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.328998182628811e-05,
+      "loss": 1.4433,
+      "step": 6580
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.326147520777118e-05,
+      "loss": 1.4082,
+      "step": 6581
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.323297538737402e-05,
+      "loss": 1.5728,
+      "step": 6582
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.320448236851117e-05,
+      "loss": 1.6551,
+      "step": 6583
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.317599615459664e-05,
+      "loss": 1.0904,
+      "step": 6584
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.314751674904343e-05,
+      "loss": 1.1619,
+      "step": 6585
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.311904415526384e-05,
+      "loss": 1.1697,
+      "step": 6586
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.309057837666929e-05,
+      "loss": 1.2622,
+      "step": 6587
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.306211941667043e-05,
+      "loss": 1.097,
+      "step": 6588
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.3033667278677056e-05,
+      "loss": 1.4471,
+      "step": 6589
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.300522196609815e-05,
+      "loss": 1.2287,
+      "step": 6590
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.297678348234185e-05,
+      "loss": 1.1944,
+      "step": 6591
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.294835183081558e-05,
+      "loss": 1.4471,
+      "step": 6592
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.29199270149258e-05,
+      "loss": 1.1325,
+      "step": 6593
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2891509038078304e-05,
+      "loss": 1.393,
+      "step": 6594
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.286309790367794e-05,
+      "loss": 1.3155,
+      "step": 6595
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.283469361512881e-05,
+      "loss": 1.0821,
+      "step": 6596
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2806296175834105e-05,
+      "loss": 1.5401,
+      "step": 6597
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2777905589196345e-05,
+      "loss": 1.3718,
+      "step": 6598
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2749521858617084e-05,
+      "loss": 1.3573,
+      "step": 6599
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2721144987497173e-05,
+      "loss": 1.3984,
+      "step": 6600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2692774979236516e-05,
+      "loss": 1.182,
+      "step": 6601
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.266441183723433e-05,
+      "loss": 1.4079,
+      "step": 6602
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.263605556488892e-05,
+      "loss": 1.1944,
+      "step": 6603
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.260770616559774e-05,
+      "loss": 1.0757,
+      "step": 6604
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.257936364275754e-05,
+      "loss": 1.3977,
+      "step": 6605
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.25510279997641e-05,
+      "loss": 1.2493,
+      "step": 6606
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2522699240012555e-05,
+      "loss": 1.527,
+      "step": 6607
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2494377366897046e-05,
+      "loss": 1.5498,
+      "step": 6608
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2466062383810956e-05,
+      "loss": 1.7054,
+      "step": 6609
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.243775429414682e-05,
+      "loss": 1.0742,
+      "step": 6610
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2409453101296425e-05,
+      "loss": 1.2293,
+      "step": 6611
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.238115880865061e-05,
+      "loss": 1.0776,
+      "step": 6612
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2352871419599535e-05,
+      "loss": 1.2114,
+      "step": 6613
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.232459093753237e-05,
+      "loss": 1.3999,
+      "step": 6614
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.22963173658376e-05,
+      "loss": 1.31,
+      "step": 6615
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.22680507079028e-05,
+      "loss": 1.2679,
+      "step": 6616
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.223979096711471e-05,
+      "loss": 1.6329,
+      "step": 6617
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.221153814685926e-05,
+      "loss": 1.3148,
+      "step": 6618
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.218329225052162e-05,
+      "loss": 1.2646,
+      "step": 6619
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2155053281486e-05,
+      "loss": 1.4544,
+      "step": 6620
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.212682124313592e-05,
+      "loss": 1.4433,
+      "step": 6621
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2098596138853955e-05,
+      "loss": 1.1002,
+      "step": 6622
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.207037797202189e-05,
+      "loss": 1.0029,
+      "step": 6623
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.204216674602066e-05,
+      "loss": 1.0971,
+      "step": 6624
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.201396246423045e-05,
+      "loss": 1.4018,
+      "step": 6625
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.19857651300305e-05,
+      "loss": 1.0918,
+      "step": 6626
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.195757474679931e-05,
+      "loss": 1.3515,
+      "step": 6627
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.192939131791446e-05,
+      "loss": 1.5519,
+      "step": 6628
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1901214846752847e-05,
+      "loss": 1.0312,
+      "step": 6629
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.187304533669028e-05,
+      "loss": 1.4387,
+      "step": 6630
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1844882791102e-05,
+      "loss": 0.9493,
+      "step": 6631
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.181672721336223e-05,
+      "loss": 1.5707,
+      "step": 6632
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1788578606844485e-05,
+      "loss": 1.0015,
+      "step": 6633
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1760436974921335e-05,
+      "loss": 1.2095,
+      "step": 6634
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1732302320964635e-05,
+      "loss": 1.3302,
+      "step": 6635
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1704174648345284e-05,
+      "loss": 1.1997,
+      "step": 6636
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1676053960433404e-05,
+      "loss": 1.1794,
+      "step": 6637
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.164794026059824e-05,
+      "loss": 1.5494,
+      "step": 6638
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1619833552208264e-05,
+      "loss": 1.3831,
+      "step": 6639
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1591733838631106e-05,
+      "loss": 1.2782,
+      "step": 6640
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.156364112323348e-05,
+      "loss": 0.931,
+      "step": 6641
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.15355554093814e-05,
+      "loss": 1.0582,
+      "step": 6642
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1507476700439804e-05,
+      "loss": 1.3176,
+      "step": 6643
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.147940499977308e-05,
+      "loss": 1.1902,
+      "step": 6644
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.145134031074452e-05,
+      "loss": 1.215,
+      "step": 6645
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.14232826367168e-05,
+      "loss": 1.0762,
+      "step": 6646
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1395231981051555e-05,
+      "loss": 1.6444,
+      "step": 6647
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.136718834710973e-05,
+      "loss": 1.0049,
+      "step": 6648
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.133915173825137e-05,
+      "loss": 1.4642,
+      "step": 6649
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.131112215783566e-05,
+      "loss": 1.2024,
+      "step": 6650
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.128309960922091e-05,
+      "loss": 1.56,
+      "step": 6651
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.125508409576473e-05,
+      "loss": 1.3235,
+      "step": 6652
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.122707562082372e-05,
+      "loss": 1.2598,
+      "step": 6653
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.119907418775378e-05,
+      "loss": 1.5899,
+      "step": 6654
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1171079799909874e-05,
+      "loss": 0.9339,
+      "step": 6655
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.114309246064613e-05,
+      "loss": 1.4496,
+      "step": 6656
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.111511217331582e-05,
+      "loss": 1.3381,
+      "step": 6657
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1087138941271453e-05,
+      "loss": 1.3387,
+      "step": 6658
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.10591727678646e-05,
+      "loss": 1.3915,
+      "step": 6659
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.103121365644608e-05,
+      "loss": 1.367,
+      "step": 6660
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.100326161036573e-05,
+      "loss": 1.0854,
+      "step": 6661
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.09753166329727e-05,
+      "loss": 0.9944,
+      "step": 6662
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.094737872761518e-05,
+      "loss": 1.4044,
+      "step": 6663
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.091944789764055e-05,
+      "loss": 1.1896,
+      "step": 6664
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.089152414639531e-05,
+      "loss": 1.3857,
+      "step": 6665
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0863607477225185e-05,
+      "loss": 1.208,
+      "step": 6666
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0835697893474955e-05,
+      "loss": 1.5027,
+      "step": 6667
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.080779539848868e-05,
+      "loss": 1.2824,
+      "step": 6668
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.077989999560945e-05,
+      "loss": 1.5031,
+      "step": 6669
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.075201168817955e-05,
+      "loss": 1.1781,
+      "step": 6670
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.072413047954038e-05,
+      "loss": 1.0882,
+      "step": 6671
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.069625637303259e-05,
+      "loss": 1.0335,
+      "step": 6672
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.066838937199585e-05,
+      "loss": 1.2192,
+      "step": 6673
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0640529479769115e-05,
+      "loss": 1.4976,
+      "step": 6674
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.061267669969033e-05,
+      "loss": 1.3278,
+      "step": 6675
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0584831035096745e-05,
+      "loss": 1.3006,
+      "step": 6676
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0556992489324675e-05,
+      "loss": 1.2139,
+      "step": 6677
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.052916106570952e-05,
+      "loss": 1.2052,
+      "step": 6678
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.050133676758598e-05,
+      "loss": 1.2473,
+      "step": 6679
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.047351959828778e-05,
+      "loss": 1.1855,
+      "step": 6680
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.044570956114786e-05,
+      "loss": 1.0162,
+      "step": 6681
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.041790665949826e-05,
+      "loss": 1.1777,
+      "step": 6682
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.039011089667019e-05,
+      "loss": 1.4251,
+      "step": 6683
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.036232227599395e-05,
+      "loss": 1.2916,
+      "step": 6684
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.033454080079909e-05,
+      "loss": 1.4928,
+      "step": 6685
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.0306766474414194e-05,
+      "loss": 1.316,
+      "step": 6686
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.02789993001671e-05,
+      "loss": 0.854,
+      "step": 6687
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.0251239281384655e-05,
+      "loss": 1.4079,
+      "step": 6688
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.022348642139303e-05,
+      "loss": 1.1078,
+      "step": 6689
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.01957407235173e-05,
+      "loss": 1.0815,
+      "step": 6690
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.016800219108191e-05,
+      "loss": 1.1092,
+      "step": 6691
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.0140270827410274e-05,
+      "loss": 1.3581,
+      "step": 6692
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.011254663582511e-05,
+      "loss": 1.0128,
+      "step": 6693
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.008482961964809e-05,
+      "loss": 1.3556,
+      "step": 6694
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.0057119782200224e-05,
+      "loss": 0.8936,
+      "step": 6695
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.00294171268015e-05,
+      "loss": 1.4647,
+      "step": 6696
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.0001721656771144e-05,
+      "loss": 1.3261,
+      "step": 6697
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.997403337542742e-05,
+      "loss": 1.4698,
+      "step": 6698
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9946352286087875e-05,
+      "loss": 1.1641,
+      "step": 6699
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9918678392069055e-05,
+      "loss": 1.3096,
+      "step": 6700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.989101169668675e-05,
+      "loss": 1.0896,
+      "step": 6701
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9863352203255846e-05,
+      "loss": 1.2864,
+      "step": 6702
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9835699915090316e-05,
+      "loss": 1.3532,
+      "step": 6703
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9808054835503314e-05,
+      "loss": 1.2036,
+      "step": 6704
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.97804169678072e-05,
+      "loss": 1.2132,
+      "step": 6705
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.975278631531332e-05,
+      "loss": 1.4303,
+      "step": 6706
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.97251628813323e-05,
+      "loss": 1.1459,
+      "step": 6707
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.969754666917378e-05,
+      "loss": 1.5405,
+      "step": 6708
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9669937682146673e-05,
+      "loss": 1.3986,
+      "step": 6709
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.964233592355889e-05,
+      "loss": 1.2184,
+      "step": 6710
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.961474139671753e-05,
+      "loss": 1.2821,
+      "step": 6711
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9587154104928856e-05,
+      "loss": 1.8392,
+      "step": 6712
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9559574051498196e-05,
+      "loss": 1.1242,
+      "step": 6713
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.95320012397301e-05,
+      "loss": 1.5498,
+      "step": 6714
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.950443567292819e-05,
+      "loss": 1.3867,
+      "step": 6715
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9476877354395216e-05,
+      "loss": 1.3001,
+      "step": 6716
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.944932628743304e-05,
+      "loss": 1.2365,
+      "step": 6717
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.942178247534276e-05,
+      "loss": 1.2954,
+      "step": 6718
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9394245921424476e-05,
+      "loss": 1.1672,
+      "step": 6719
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.936671662897753e-05,
+      "loss": 1.0085,
+      "step": 6720
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9339194601300274e-05,
+      "loss": 0.9968,
+      "step": 6721
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.931167984169035e-05,
+      "loss": 1.4968,
+      "step": 6722
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.928417235344437e-05,
+      "loss": 1.2828,
+      "step": 6723
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.925667213985815e-05,
+      "loss": 1.4344,
+      "step": 6724
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9229179204226606e-05,
+      "loss": 1.1712,
+      "step": 6725
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9201693549843843e-05,
+      "loss": 1.2343,
+      "step": 6726
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9174215180003e-05,
+      "loss": 1.4183,
+      "step": 6727
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.914674409799647e-05,
+      "loss": 1.3225,
+      "step": 6728
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.911928030711565e-05,
+      "loss": 1.4351,
+      "step": 6729
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.909182381065111e-05,
+      "loss": 1.2144,
+      "step": 6730
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.906437461189251e-05,
+      "loss": 1.0913,
+      "step": 6731
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.903693271412875e-05,
+      "loss": 1.4895,
+      "step": 6732
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.9009498120647715e-05,
+      "loss": 1.0788,
+      "step": 6733
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.898207083473653e-05,
+      "loss": 1.1113,
+      "step": 6734
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.895465085968134e-05,
+      "loss": 0.968,
+      "step": 6735
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8927238198767516e-05,
+      "loss": 1.1546,
+      "step": 6736
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.889983285527948e-05,
+      "loss": 0.7417,
+      "step": 6737
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8872434832500807e-05,
+      "loss": 1.3888,
+      "step": 6738
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.884504413371414e-05,
+      "loss": 1.5972,
+      "step": 6739
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.881766076220137e-05,
+      "loss": 1.4026,
+      "step": 6740
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8790284721243354e-05,
+      "loss": 0.848,
+      "step": 6741
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.876291601412022e-05,
+      "loss": 1.1906,
+      "step": 6742
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8735554644111124e-05,
+      "loss": 1.2563,
+      "step": 6743
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.870820061449436e-05,
+      "loss": 1.4028,
+      "step": 6744
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.86808539285473e-05,
+      "loss": 1.5027,
+      "step": 6745
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.865351458954658e-05,
+      "loss": 1.2345,
+      "step": 6746
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.862618260076776e-05,
+      "loss": 1.3114,
+      "step": 6747
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.859885796548567e-05,
+      "loss": 1.8188,
+      "step": 6748
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.857154068697427e-05,
+      "loss": 1.3693,
+      "step": 6749
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.854423076850645e-05,
+      "loss": 1.3002,
+      "step": 6750
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.851692821335444e-05,
+      "loss": 1.3283,
+      "step": 6751
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.848963302478943e-05,
+      "loss": 1.1397,
+      "step": 6752
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.846234520608185e-05,
+      "loss": 1.2632,
+      "step": 6753
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8435064760501114e-05,
+      "loss": 1.4652,
+      "step": 6754
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.840779169131592e-05,
+      "loss": 1.415,
+      "step": 6755
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8380526001793915e-05,
+      "loss": 1.1192,
+      "step": 6756
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.835326769520197e-05,
+      "loss": 1.4855,
+      "step": 6757
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8326016774805976e-05,
+      "loss": 1.3695,
+      "step": 6758
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.829877324387108e-05,
+      "loss": 1.3372,
+      "step": 6759
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.827153710566139e-05,
+      "loss": 1.2242,
+      "step": 6760
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.824430836344027e-05,
+      "loss": 1.3112,
+      "step": 6761
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.82170870204701e-05,
+      "loss": 1.4232,
+      "step": 6762
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.81898730800124e-05,
+      "loss": 1.3206,
+      "step": 6763
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.816266654532775e-05,
+      "loss": 1.0496,
+      "step": 6764
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8135467419675997e-05,
+      "loss": 1.425,
+      "step": 6765
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.810827570631591e-05,
+      "loss": 1.4287,
+      "step": 6766
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8081091408505545e-05,
+      "loss": 1.2706,
+      "step": 6767
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.805391452950191e-05,
+      "loss": 1.272,
+      "step": 6768
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.802674507256128e-05,
+      "loss": 1.2867,
+      "step": 6769
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.799958304093889e-05,
+      "loss": 1.2233,
+      "step": 6770
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.79724284378892e-05,
+      "loss": 1.4,
+      "step": 6771
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7945281266665676e-05,
+      "loss": 1.4155,
+      "step": 6772
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7918141530521025e-05,
+      "loss": 1.4098,
+      "step": 6773
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.789100923270693e-05,
+      "loss": 1.3791,
+      "step": 6774
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7863884376474315e-05,
+      "loss": 1.2209,
+      "step": 6775
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.783676696507308e-05,
+      "loss": 1.4295,
+      "step": 6776
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.780965700175233e-05,
+      "loss": 1.0888,
+      "step": 6777
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7782554489760183e-05,
+      "loss": 1.8336,
+      "step": 6778
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.775545943234401e-05,
+      "loss": 1.1288,
+      "step": 6779
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.772837183275012e-05,
+      "loss": 1.1346,
+      "step": 6780
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.770129169422408e-05,
+      "loss": 1.3763,
+      "step": 6781
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.767421902001044e-05,
+      "loss": 1.6343,
+      "step": 6782
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.764715381335295e-05,
+      "loss": 1.5923,
+      "step": 6783
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.762009607749441e-05,
+      "loss": 1.1929,
+      "step": 6784
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7593045815676716e-05,
+      "loss": 0.9791,
+      "step": 6785
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7566003031140937e-05,
+      "loss": 1.0774,
+      "step": 6786
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.753896772712715e-05,
+      "loss": 1.2676,
+      "step": 6787
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7511939906874635e-05,
+      "loss": 1.3101,
+      "step": 6788
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7484919573621726e-05,
+      "loss": 1.2883,
+      "step": 6789
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.745790673060582e-05,
+      "loss": 1.4098,
+      "step": 6790
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.743090138106346e-05,
+      "loss": 1.2633,
+      "step": 6791
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.740390352823032e-05,
+      "loss": 1.1992,
+      "step": 6792
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.737691317534111e-05,
+      "loss": 1.3912,
+      "step": 6793
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7349930325629736e-05,
+      "loss": 1.147,
+      "step": 6794
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.732295498232906e-05,
+      "loss": 1.3033,
+      "step": 6795
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.729598714867122e-05,
+      "loss": 1.279,
+      "step": 6796
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.726902682788731e-05,
+      "loss": 1.3551,
+      "step": 6797
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7242074023207594e-05,
+      "loss": 1.1651,
+      "step": 6798
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7215128737861383e-05,
+      "loss": 1.3249,
+      "step": 6799
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.718819097507717e-05,
+      "loss": 1.4651,
+      "step": 6800
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.716126073808247e-05,
+      "loss": 1.2947,
+      "step": 6801
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.713433803010396e-05,
+      "loss": 1.002,
+      "step": 6802
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.710742285436737e-05,
+      "loss": 1.348,
+      "step": 6803
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.708051521409751e-05,
+      "loss": 1.3851,
+      "step": 6804
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.705361511251831e-05,
+      "loss": 1.6274,
+      "step": 6805
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7026722552852864e-05,
+      "loss": 1.4162,
+      "step": 6806
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.699983753832324e-05,
+      "loss": 1.2307,
+      "step": 6807
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.69729600721507e-05,
+      "loss": 1.499,
+      "step": 6808
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6946090157555556e-05,
+      "loss": 1.3185,
+      "step": 6809
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.691922779775723e-05,
+      "loss": 1.7548,
+      "step": 6810
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.689237299597418e-05,
+      "loss": 1.502,
+      "step": 6811
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.686552575542408e-05,
+      "loss": 0.8052,
+      "step": 6812
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.683868607932357e-05,
+      "loss": 1.4685,
+      "step": 6813
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.681185397088851e-05,
+      "loss": 1.2744,
+      "step": 6814
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6785029433333715e-05,
+      "loss": 1.3964,
+      "step": 6815
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6758212469873234e-05,
+      "loss": 1.4484,
+      "step": 6816
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.673140308372009e-05,
+      "loss": 1.0482,
+      "step": 6817
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6704601278086484e-05,
+      "loss": 1.4272,
+      "step": 6818
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.66778070561836e-05,
+      "loss": 1.1213,
+      "step": 6819
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.665102042122183e-05,
+      "loss": 1.1909,
+      "step": 6820
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.662424137641064e-05,
+      "loss": 1.2813,
+      "step": 6821
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.659746992495854e-05,
+      "loss": 1.2693,
+      "step": 6822
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6570706070073133e-05,
+      "loss": 1.4081,
+      "step": 6823
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.65439498149611e-05,
+      "loss": 1.2999,
+      "step": 6824
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.651720116282831e-05,
+      "loss": 1.2417,
+      "step": 6825
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6490460116879565e-05,
+      "loss": 1.3164,
+      "step": 6826
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.646372668031893e-05,
+      "loss": 1.3366,
+      "step": 6827
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.64370008563494e-05,
+      "loss": 1.3539,
+      "step": 6828
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6410282648173176e-05,
+      "loss": 1.598,
+      "step": 6829
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6383572058991466e-05,
+      "loss": 1.5388,
+      "step": 6830
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.635686909200461e-05,
+      "loss": 1.3742,
+      "step": 6831
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.633017375041198e-05,
+      "loss": 1.3156,
+      "step": 6832
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6303486037412135e-05,
+      "loss": 1.1622,
+      "step": 6833
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.627680595620261e-05,
+      "loss": 1.1045,
+      "step": 6834
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.625013350998013e-05,
+      "loss": 1.5774,
+      "step": 6835
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.622346870194042e-05,
+      "loss": 1.3754,
+      "step": 6836
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.619681153527832e-05,
+      "loss": 1.1602,
+      "step": 6837
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6170162013187724e-05,
+      "loss": 0.9688,
+      "step": 6838
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6143520138861706e-05,
+      "loss": 0.9914,
+      "step": 6839
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6116885915492304e-05,
+      "loss": 1.2579,
+      "step": 6840
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.609025934627074e-05,
+      "loss": 1.4717,
+      "step": 6841
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.606364043438723e-05,
+      "loss": 1.313,
+      "step": 6842
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6037029183031155e-05,
+      "loss": 0.9684,
+      "step": 6843
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.601042559539093e-05,
+      "loss": 1.4365,
+      "step": 6844
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.598382967465404e-05,
+      "loss": 1.305,
+      "step": 6845
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.595724142400706e-05,
+      "loss": 1.3063,
+      "step": 6846
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.593066084663571e-05,
+      "loss": 1.0918,
+      "step": 6847
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5904087945724685e-05,
+      "loss": 1.2309,
+      "step": 6848
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.587752272445787e-05,
+      "loss": 1.3588,
+      "step": 6849
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5850965186018136e-05,
+      "loss": 1.0677,
+      "step": 6850
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.582441533358748e-05,
+      "loss": 1.4779,
+      "step": 6851
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.579787317034694e-05,
+      "loss": 1.4329,
+      "step": 6852
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.577133869947672e-05,
+      "loss": 1.4602,
+      "step": 6853
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.574481192415598e-05,
+      "loss": 1.0438,
+      "step": 6854
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.571829284756306e-05,
+      "loss": 1.3749,
+      "step": 6855
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.569178147287537e-05,
+      "loss": 1.2111,
+      "step": 6856
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.566527780326934e-05,
+      "loss": 1.2616,
+      "step": 6857
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.563878184192049e-05,
+      "loss": 1.2396,
+      "step": 6858
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.561229359200341e-05,
+      "loss": 1.0762,
+      "step": 6859
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5585813056691844e-05,
+      "loss": 1.1346,
+      "step": 6860
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.55593402391585e-05,
+      "loss": 1.029,
+      "step": 6861
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.553287514257526e-05,
+      "loss": 1.4138,
+      "step": 6862
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5506417770113034e-05,
+      "loss": 1.3687,
+      "step": 6863
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.547996812494179e-05,
+      "loss": 0.8122,
+      "step": 6864
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.545352621023056e-05,
+      "loss": 1.559,
+      "step": 6865
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.542709202914753e-05,
+      "loss": 1.4162,
+      "step": 6866
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.540066558485987e-05,
+      "loss": 1.3452,
+      "step": 6867
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.537424688053391e-05,
+      "loss": 0.885,
+      "step": 6868
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.534783591933497e-05,
+      "loss": 1.7299,
+      "step": 6869
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.532143270442748e-05,
+      "loss": 1.3892,
+      "step": 6870
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.529503723897492e-05,
+      "loss": 1.2022,
+      "step": 6871
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.52686495261399e-05,
+      "loss": 1.2452,
+      "step": 6872
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.524226956908399e-05,
+      "loss": 0.8989,
+      "step": 6873
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.521589737096801e-05,
+      "loss": 0.9981,
+      "step": 6874
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.518953293495164e-05,
+      "loss": 1.3051,
+      "step": 6875
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.516317626419381e-05,
+      "loss": 1.4471,
+      "step": 6876
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.51368273618524e-05,
+      "loss": 1.075,
+      "step": 6877
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.51104862310844e-05,
+      "loss": 1.6901,
+      "step": 6878
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.508415287504586e-05,
+      "loss": 1.5249,
+      "step": 6879
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.505782729689194e-05,
+      "loss": 1.1745,
+      "step": 6880
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5031509499776804e-05,
+      "loss": 1.1971,
+      "step": 6881
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.500519948685375e-05,
+      "loss": 1.2055,
+      "step": 6882
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.497889726127509e-05,
+      "loss": 1.6324,
+      "step": 6883
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.495260282619224e-05,
+      "loss": 1.4923,
+      "step": 6884
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4926316184755594e-05,
+      "loss": 1.2249,
+      "step": 6885
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.490003734011478e-05,
+      "loss": 1.8758,
+      "step": 6886
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.487376629541831e-05,
+      "loss": 1.7079,
+      "step": 6887
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4847503053813924e-05,
+      "loss": 1.179,
+      "step": 6888
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4821247618448286e-05,
+      "loss": 1.4967,
+      "step": 6889
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4794999992467236e-05,
+      "loss": 1.5529,
+      "step": 6890
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.476876017901561e-05,
+      "loss": 1.2714,
+      "step": 6891
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.47425281812373e-05,
+      "loss": 1.535,
+      "step": 6892
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.471630400227534e-05,
+      "loss": 1.0558,
+      "step": 6893
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.469008764527173e-05,
+      "loss": 1.0605,
+      "step": 6894
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.466387911336762e-05,
+      "loss": 1.1762,
+      "step": 6895
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.463767840970319e-05,
+      "loss": 1.2189,
+      "step": 6896
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.461148553741763e-05,
+      "loss": 1.1833,
+      "step": 6897
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4585300499649234e-05,
+      "loss": 1.3967,
+      "step": 6898
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.45591232995354e-05,
+      "loss": 1.0951,
+      "step": 6899
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4532953940212495e-05,
+      "loss": 0.9736,
+      "step": 6900
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.450679242481607e-05,
+      "loss": 1.5221,
+      "step": 6901
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4480638756480586e-05,
+      "loss": 1.2075,
+      "step": 6902
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.445449293833972e-05,
+      "loss": 1.0247,
+      "step": 6903
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.442835497352608e-05,
+      "loss": 1.3808,
+      "step": 6904
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.44022248651714e-05,
+      "loss": 1.5751,
+      "step": 6905
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4376102616406404e-05,
+      "loss": 1.976,
+      "step": 6906
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.434998823036101e-05,
+      "loss": 1.2926,
+      "step": 6907
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4323881710164044e-05,
+      "loss": 1.3348,
+      "step": 6908
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.42977830589435e-05,
+      "loss": 0.8955,
+      "step": 6909
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.427169227982637e-05,
+      "loss": 1.373,
+      "step": 6910
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.424560937593872e-05,
+      "loss": 1.1603,
+      "step": 6911
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.421953435040564e-05,
+      "loss": 1.0795,
+      "step": 6912
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.419346720635135e-05,
+      "loss": 1.4664,
+      "step": 6913
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.416740794689903e-05,
+      "loss": 1.6519,
+      "step": 6914
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.414135657517103e-05,
+      "loss": 1.2706,
+      "step": 6915
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.4115313094288635e-05,
+      "loss": 1.5101,
+      "step": 6916
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.408927750737234e-05,
+      "loss": 1.4528,
+      "step": 6917
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.406324981754145e-05,
+      "loss": 1.3186,
+      "step": 6918
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.403723002791458e-05,
+      "loss": 1.0286,
+      "step": 6919
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.401121814160921e-05,
+      "loss": 1.2013,
+      "step": 6920
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.398521416174204e-05,
+      "loss": 1.3839,
+      "step": 6921
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.395921809142865e-05,
+      "loss": 1.1666,
+      "step": 6922
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3933229933783814e-05,
+      "loss": 1.2423,
+      "step": 6923
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3907249691921286e-05,
+      "loss": 1.3978,
+      "step": 6924
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.388127736895388e-05,
+      "loss": 1.0567,
+      "step": 6925
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.385531296799343e-05,
+      "loss": 1.1798,
+      "step": 6926
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.382935649215089e-05,
+      "loss": 1.2833,
+      "step": 6927
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3803407944536267e-05,
+      "loss": 1.4236,
+      "step": 6928
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.377746732825856e-05,
+      "loss": 1.4519,
+      "step": 6929
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.375153464642583e-05,
+      "loss": 1.3802,
+      "step": 6930
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.372560990214516e-05,
+      "loss": 1.5227,
+      "step": 6931
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.36996930985228e-05,
+      "loss": 1.2185,
+      "step": 6932
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.36737842386639e-05,
+      "loss": 1.3635,
+      "step": 6933
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.364788332567278e-05,
+      "loss": 1.2968,
+      "step": 6934
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.36219903626527e-05,
+      "loss": 1.7109,
+      "step": 6935
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.359610535270609e-05,
+      "loss": 1.3373,
+      "step": 6936
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3570228298934315e-05,
+      "loss": 1.2685,
+      "step": 6937
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3544359204437844e-05,
+      "loss": 1.2278,
+      "step": 6938
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.351849807231614e-05,
+      "loss": 1.3914,
+      "step": 6939
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3492644905667826e-05,
+      "loss": 1.4001,
+      "step": 6940
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.346679970759041e-05,
+      "loss": 1.2771,
+      "step": 6941
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3440962481180616e-05,
+      "loss": 1.3187,
+      "step": 6942
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.34151332295341e-05,
+      "loss": 1.5406,
+      "step": 6943
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.338931195574557e-05,
+      "loss": 1.4418,
+      "step": 6944
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3363498662908775e-05,
+      "loss": 1.449,
+      "step": 6945
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.33376933541166e-05,
+      "loss": 1.1994,
+      "step": 6946
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.331189603246084e-05,
+      "loss": 1.1993,
+      "step": 6947
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.328610670103247e-05,
+      "loss": 1.0699,
+      "step": 6948
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3260325362921355e-05,
+      "loss": 1.4796,
+      "step": 6949
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.323455202121656e-05,
+      "loss": 1.2401,
+      "step": 6950
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.320878667900607e-05,
+      "loss": 1.2952,
+      "step": 6951
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.318302933937697e-05,
+      "loss": 1.2583,
+      "step": 6952
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3157280005415326e-05,
+      "loss": 1.5283,
+      "step": 6953
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3131538680206376e-05,
+      "loss": 1.156,
+      "step": 6954
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.310580536683423e-05,
+      "loss": 1.2802,
+      "step": 6955
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.308008006838219e-05,
+      "loss": 0.9566,
+      "step": 6956
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3054362787932513e-05,
+      "loss": 1.0619,
+      "step": 6957
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.30286535285665e-05,
+      "loss": 1.4315,
+      "step": 6958
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3002952293364465e-05,
+      "loss": 1.4421,
+      "step": 6959
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2977259085405866e-05,
+      "loss": 1.4137,
+      "step": 6960
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.295157390776908e-05,
+      "loss": 1.1756,
+      "step": 6961
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2925896763531624e-05,
+      "loss": 0.8745,
+      "step": 6962
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.290022765576994e-05,
+      "loss": 1.2254,
+      "step": 6963
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.287456658755964e-05,
+      "loss": 1.3041,
+      "step": 6964
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2848913561975256e-05,
+      "loss": 1.4657,
+      "step": 6965
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.282326858209039e-05,
+      "loss": 1.238,
+      "step": 6966
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.279763165097774e-05,
+      "loss": 1.4205,
+      "step": 6967
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2772002771708946e-05,
+      "loss": 1.2895,
+      "step": 6968
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.274638194735477e-05,
+      "loss": 1.137,
+      "step": 6969
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.272076918098494e-05,
+      "loss": 1.2296,
+      "step": 6970
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.269516447566826e-05,
+      "loss": 1.353,
+      "step": 6971
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2669567834472526e-05,
+      "loss": 1.2001,
+      "step": 6972
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.264397926046464e-05,
+      "loss": 1.1757,
+      "step": 6973
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.261839875671044e-05,
+      "loss": 1.4509,
+      "step": 6974
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.259282632627492e-05,
+      "loss": 1.1781,
+      "step": 6975
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2567261972221966e-05,
+      "loss": 1.2589,
+      "step": 6976
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.254170569761467e-05,
+      "loss": 1.0264,
+      "step": 6977
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.251615750551492e-05,
+      "loss": 1.4884,
+      "step": 6978
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.249061739898388e-05,
+      "loss": 1.2389,
+      "step": 6979
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2465085381081554e-05,
+      "loss": 1.2031,
+      "step": 6980
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.243956145486713e-05,
+      "loss": 1.2034,
+      "step": 6981
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.241404562339868e-05,
+      "loss": 1.4544,
+      "step": 6982
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.238853788973346e-05,
+      "loss": 1.3363,
+      "step": 6983
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2363038256927636e-05,
+      "loss": 1.3669,
+      "step": 6984
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.233754672803644e-05,
+      "loss": 1.144,
+      "step": 6985
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2312063306114106e-05,
+      "loss": 1.562,
+      "step": 6986
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2286587994214e-05,
+      "loss": 0.882,
+      "step": 6987
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2261120795388376e-05,
+      "loss": 1.2646,
+      "step": 6988
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2235661712688636e-05,
+      "loss": 1.4291,
+      "step": 6989
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.221021074916514e-05,
+      "loss": 1.3789,
+      "step": 6990
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.218476790786729e-05,
+      "loss": 1.4787,
+      "step": 6991
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2159333191843475e-05,
+      "loss": 1.1776,
+      "step": 6992
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.213390660414122e-05,
+      "loss": 1.3813,
+      "step": 6993
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.210848814780695e-05,
+      "loss": 1.2143,
+      "step": 6994
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.208307782588624e-05,
+      "loss": 1.2845,
+      "step": 6995
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2057675641423554e-05,
+      "loss": 1.176,
+      "step": 6996
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.203228159746251e-05,
+      "loss": 1.4114,
+      "step": 6997
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.200689569704568e-05,
+      "loss": 1.549,
+      "step": 6998
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.198151794321462e-05,
+      "loss": 1.3119,
+      "step": 6999
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.195614833901005e-05,
+      "loss": 1.4873,
+      "step": 7000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.193078688747153e-05,
+      "loss": 1.5131,
+      "step": 7001
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1905433591637834e-05,
+      "loss": 1.5328,
+      "step": 7002
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.188008845454661e-05,
+      "loss": 1.17,
+      "step": 7003
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1854751479234604e-05,
+      "loss": 1.509,
+      "step": 7004
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.182942266873751e-05,
+      "loss": 1.0964,
+      "step": 7005
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.180410202609017e-05,
+      "loss": 1.3486,
+      "step": 7006
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1778789554326304e-05,
+      "loss": 1.2287,
+      "step": 7007
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.17534852564788e-05,
+      "loss": 1.1395,
+      "step": 7008
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.172818913557941e-05,
+      "loss": 1.3247,
+      "step": 7009
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1702901194659075e-05,
+      "loss": 1.5786,
+      "step": 7010
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1677621436747596e-05,
+      "loss": 1.1774,
+      "step": 7011
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1652349864873896e-05,
+      "loss": 1.2312,
+      "step": 7012
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.162708648206584e-05,
+      "loss": 1.2737,
+      "step": 7013
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1601831291350424e-05,
+      "loss": 1.4678,
+      "step": 7014
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.157658429575353e-05,
+      "loss": 1.3008,
+      "step": 7015
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1551345498300186e-05,
+      "loss": 1.6668,
+      "step": 7016
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1526114902014345e-05,
+      "loss": 1.3527,
+      "step": 7017
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1500892509919025e-05,
+      "loss": 1.2745,
+      "step": 7018
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.147567832503618e-05,
+      "loss": 1.2819,
+      "step": 7019
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.145047235038694e-05,
+      "loss": 1.3665,
+      "step": 7020
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.142527458899126e-05,
+      "loss": 1.3433,
+      "step": 7021
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.14000850438683e-05,
+      "loss": 1.2569,
+      "step": 7022
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.137490371803606e-05,
+      "loss": 1.1505,
+      "step": 7023
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.134973061451174e-05,
+      "loss": 0.9973,
+      "step": 7024
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.132456573631133e-05,
+      "loss": 1.2599,
+      "step": 7025
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.129940908645005e-05,
+      "loss": 1.5982,
+      "step": 7026
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.127426066794197e-05,
+      "loss": 1.1799,
+      "step": 7027
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.124912048380031e-05,
+      "loss": 1.3293,
+      "step": 7028
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1223988537037176e-05,
+      "loss": 1.2576,
+      "step": 7029
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1198864830663824e-05,
+      "loss": 1.3232,
+      "step": 7030
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.11737493676904e-05,
+      "loss": 1.0993,
+      "step": 7031
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.114864215112612e-05,
+      "loss": 1.2878,
+      "step": 7032
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.112354318397915e-05,
+      "loss": 0.8956,
+      "step": 7033
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1098452469256825e-05,
+      "loss": 1.4776,
+      "step": 7034
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1073370009965286e-05,
+      "loss": 1.3063,
+      "step": 7035
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.104829580910983e-05,
+      "loss": 1.3007,
+      "step": 7036
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.102322986969478e-05,
+      "loss": 1.3814,
+      "step": 7037
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0998172194723294e-05,
+      "loss": 1.366,
+      "step": 7038
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.097312278719773e-05,
+      "loss": 1.0363,
+      "step": 7039
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.094808165011932e-05,
+      "loss": 1.2047,
+      "step": 7040
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.092304878648844e-05,
+      "loss": 1.1424,
+      "step": 7041
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.089802419930433e-05,
+      "loss": 1.3089,
+      "step": 7042
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.087300789156537e-05,
+      "loss": 1.466,
+      "step": 7043
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.084799986626886e-05,
+      "loss": 1.0916,
+      "step": 7044
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.082300012641113e-05,
+      "loss": 1.395,
+      "step": 7045
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.079800867498748e-05,
+      "loss": 1.503,
+      "step": 7046
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.077302551499234e-05,
+      "loss": 1.5688,
+      "step": 7047
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.074805064941898e-05,
+      "loss": 1.4183,
+      "step": 7048
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0723084081259835e-05,
+      "loss": 0.9872,
+      "step": 7049
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.069812581350624e-05,
+      "loss": 1.618,
+      "step": 7050
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.067317584914857e-05,
+      "loss": 1.5967,
+      "step": 7051
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0648234191176165e-05,
+      "loss": 1.1997,
+      "step": 7052
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0623300842577475e-05,
+      "loss": 1.6217,
+      "step": 7053
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.059837580633981e-05,
+      "loss": 1.4565,
+      "step": 7054
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.057345908544964e-05,
+      "loss": 1.0925,
+      "step": 7055
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0548550682892284e-05,
+      "loss": 1.1045,
+      "step": 7056
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.052365060165222e-05,
+      "loss": 1.2053,
+      "step": 7057
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0498758844712794e-05,
+      "loss": 0.8976,
+      "step": 7058
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.047387541505642e-05,
+      "loss": 1.0902,
+      "step": 7059
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0449000315664466e-05,
+      "loss": 1.5832,
+      "step": 7060
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0424133549517418e-05,
+      "loss": 1.389,
+      "step": 7061
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0399275119594605e-05,
+      "loss": 1.1594,
+      "step": 7062
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.037442502887451e-05,
+      "loss": 1.6234,
+      "step": 7063
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0349583280334504e-05,
+      "loss": 1.6499,
+      "step": 7064
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0324749876950997e-05,
+      "loss": 1.2215,
+      "step": 7065
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0299924821699378e-05,
+      "loss": 1.4264,
+      "step": 7066
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.027510811755412e-05,
+      "loss": 1.5749,
+      "step": 7067
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0250299767488555e-05,
+      "loss": 1.2064,
+      "step": 7068
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0225499774475174e-05,
+      "loss": 1.2677,
+      "step": 7069
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0200708141485323e-05,
+      "loss": 0.9771,
+      "step": 7070
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0175924871489457e-05,
+      "loss": 1.0277,
+      "step": 7071
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.015114996745696e-05,
+      "loss": 1.2609,
+      "step": 7072
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0126383432356198e-05,
+      "loss": 1.399,
+      "step": 7073
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0101625269154644e-05,
+      "loss": 1.1442,
+      "step": 7074
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.007687548081861e-05,
+      "loss": 1.6083,
+      "step": 7075
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0052134070313563e-05,
+      "loss": 1.0421,
+      "step": 7076
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0027401040603875e-05,
+      "loss": 1.4572,
+      "step": 7077
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0002676394652905e-05,
+      "loss": 1.0375,
+      "step": 7078
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.997796013542302e-05,
+      "loss": 1.1249,
+      "step": 7079
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9953252265875665e-05,
+      "loss": 1.272,
+      "step": 7080
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.992855278897112e-05,
+      "loss": 1.4194,
+      "step": 7081
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9903861707668834e-05,
+      "loss": 1.4057,
+      "step": 7082
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9879179024927095e-05,
+      "loss": 1.3738,
+      "step": 7083
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.985450474370336e-05,
+      "loss": 1.4006,
+      "step": 7084
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.982983886695384e-05,
+      "loss": 1.4025,
+      "step": 7085
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9805181397633974e-05,
+      "loss": 1.3228,
+      "step": 7086
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.978053233869803e-05,
+      "loss": 0.9922,
+      "step": 7087
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.975589169309938e-05,
+      "loss": 1.3239,
+      "step": 7088
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9731259463790307e-05,
+      "loss": 1.0777,
+      "step": 7089
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9706635653722146e-05,
+      "loss": 0.9277,
+      "step": 7090
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.96820202658452e-05,
+      "loss": 1.2998,
+      "step": 7091
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.965741330310874e-05,
+      "loss": 1.2347,
+      "step": 7092
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9632814768461002e-05,
+      "loss": 1.259,
+      "step": 7093
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9608224664849348e-05,
+      "loss": 1.0667,
+      "step": 7094
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9583642995219963e-05,
+      "loss": 1.2987,
+      "step": 7095
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9559069762518156e-05,
+      "loss": 1.3261,
+      "step": 7096
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9534504969688137e-05,
+      "loss": 1.1829,
+      "step": 7097
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9509948619673134e-05,
+      "loss": 1.1404,
+      "step": 7098
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9485400715415324e-05,
+      "loss": 1.4371,
+      "step": 7099
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9460861259855987e-05,
+      "loss": 1.4589,
+      "step": 7100
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.943633025593525e-05,
+      "loss": 1.3977,
+      "step": 7101
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9411807706592355e-05,
+      "loss": 1.3401,
+      "step": 7102
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.93872936147654e-05,
+      "loss": 1.4237,
+      "step": 7103
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.936278798339159e-05,
+      "loss": 1.3327,
+      "step": 7104
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9338290815407067e-05,
+      "loss": 1.3949,
+      "step": 7105
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9313802113746925e-05,
+      "loss": 1.387,
+      "step": 7106
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9289321881345254e-05,
+      "loss": 1.1283,
+      "step": 7107
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.926485012113519e-05,
+      "loss": 1.4851,
+      "step": 7108
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9240386836048828e-05,
+      "loss": 1.7435,
+      "step": 7109
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.921593202901721e-05,
+      "loss": 1.2643,
+      "step": 7110
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.91914857029704e-05,
+      "loss": 1.1702,
+      "step": 7111
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.916704786083738e-05,
+      "loss": 1.3281,
+      "step": 7112
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9142618505546238e-05,
+      "loss": 1.3747,
+      "step": 7113
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9118197640023902e-05,
+      "loss": 1.5637,
+      "step": 7114
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.909378526719644e-05,
+      "loss": 1.1704,
+      "step": 7115
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9069381389988736e-05,
+      "loss": 1.3525,
+      "step": 7116
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9044986011324816e-05,
+      "loss": 1.2799,
+      "step": 7117
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9020599134127557e-05,
+      "loss": 1.5628,
+      "step": 7118
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.899622076131888e-05,
+      "loss": 0.8715,
+      "step": 7119
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8971850895819663e-05,
+      "loss": 1.0186,
+      "step": 7120
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.894748954054981e-05,
+      "loss": 1.186,
+      "step": 7121
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.892313669842813e-05,
+      "loss": 1.1477,
+      "step": 7122
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.889879237237252e-05,
+      "loss": 1.2738,
+      "step": 7123
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8874456565299746e-05,
+      "loss": 1.3746,
+      "step": 7124
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8850129280125616e-05,
+      "loss": 1.123,
+      "step": 7125
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8825810519764863e-05,
+      "loss": 1.2557,
+      "step": 7126
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.88015002871313e-05,
+      "loss": 0.9959,
+      "step": 7127
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8777198585137598e-05,
+      "loss": 1.3496,
+      "step": 7128
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8752905416695508e-05,
+      "loss": 1.2514,
+      "step": 7129
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8728620784715665e-05,
+      "loss": 1.6463,
+      "step": 7130
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8704344692107822e-05,
+      "loss": 1.1923,
+      "step": 7131
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.868007714178048e-05,
+      "loss": 1.0574,
+      "step": 7132
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8655818136641366e-05,
+      "loss": 1.4646,
+      "step": 7133
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8631567679596992e-05,
+      "loss": 1.0901,
+      "step": 7134
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8607325773552996e-05,
+      "loss": 1.1492,
+      "step": 7135
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.858309242141385e-05,
+      "loss": 1.2464,
+      "step": 7136
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.855886762608313e-05,
+      "loss": 1.3265,
+      "step": 7137
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8534651390463295e-05,
+      "loss": 1.1358,
+      "step": 7138
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8510443717455814e-05,
+      "loss": 1.6112,
+      "step": 7139
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.848624460996111e-05,
+      "loss": 1.2455,
+      "step": 7140
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.846205407087864e-05,
+      "loss": 1.2171,
+      "step": 7141
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8437872103106723e-05,
+      "loss": 1.0887,
+      "step": 7142
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8413698709542803e-05,
+      "loss": 1.3148,
+      "step": 7143
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8389533893083164e-05,
+      "loss": 1.3301,
+      "step": 7144
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8365377656623093e-05,
+      "loss": 1.3149,
+      "step": 7145
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8341230003056906e-05,
+      "loss": 1.2143,
+      "step": 7146
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8317090935277812e-05,
+      "loss": 1.2452,
+      "step": 7147
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8292960456178087e-05,
+      "loss": 1.3269,
+      "step": 7148
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8268838568648857e-05,
+      "loss": 1.3867,
+      "step": 7149
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8244725275580342e-05,
+      "loss": 1.3962,
+      "step": 7150
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8220620579861646e-05,
+      "loss": 1.2684,
+      "step": 7151
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8196524484380872e-05,
+      "loss": 1.0991,
+      "step": 7152
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8172436992025066e-05,
+      "loss": 1.4616,
+      "step": 7153
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8148358105680304e-05,
+      "loss": 1.1723,
+      "step": 7154
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8124287828231566e-05,
+      "loss": 1.482,
+      "step": 7155
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8100226162562858e-05,
+      "loss": 1.2164,
+      "step": 7156
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8076173111557136e-05,
+      "loss": 1.1596,
+      "step": 7157
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8052128678096267e-05,
+      "loss": 1.3041,
+      "step": 7158
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.802809286506114e-05,
+      "loss": 1.6446,
+      "step": 7159
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8004065675331647e-05,
+      "loss": 1.0962,
+      "step": 7160
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7980047111786545e-05,
+      "loss": 1.2332,
+      "step": 7161
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7956037177303673e-05,
+      "loss": 1.3278,
+      "step": 7162
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.793203587475971e-05,
+      "loss": 1.5164,
+      "step": 7163
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7908043207030453e-05,
+      "loss": 1.2779,
+      "step": 7164
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.788405917699053e-05,
+      "loss": 1.1393,
+      "step": 7165
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7860083787513604e-05,
+      "loss": 1.3246,
+      "step": 7166
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7836117041472233e-05,
+      "loss": 1.2035,
+      "step": 7167
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7812158941738054e-05,
+      "loss": 1.4625,
+      "step": 7168
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.778820949118155e-05,
+      "loss": 1.5962,
+      "step": 7169
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.776426869267228e-05,
+      "loss": 1.444,
+      "step": 7170
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7740336549078684e-05,
+      "loss": 1.3809,
+      "step": 7171
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7716413063268177e-05,
+      "loss": 1.4403,
+      "step": 7172
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7692498238107122e-05,
+      "loss": 1.3466,
+      "step": 7173
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7668592076460932e-05,
+      "loss": 1.2485,
+      "step": 7174
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7644694581193862e-05,
+      "loss": 1.7569,
+      "step": 7175
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7620805755169255e-05,
+      "loss": 1.1527,
+      "step": 7176
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.759692560124928e-05,
+      "loss": 1.2965,
+      "step": 7177
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7573054122295226e-05,
+      "loss": 1.497,
+      "step": 7178
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7549191321167124e-05,
+      "loss": 0.8523,
+      "step": 7179
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7525337200724166e-05,
+      "loss": 1.1771,
+      "step": 7180
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.750149176382445e-05,
+      "loss": 1.5964,
+      "step": 7181
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7477655013324966e-05,
+      "loss": 1.0432,
+      "step": 7182
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7453826952081752e-05,
+      "loss": 1.5687,
+      "step": 7183
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7430007582949757e-05,
+      "loss": 1.1608,
+      "step": 7184
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7406196908782877e-05,
+      "loss": 1.4798,
+      "step": 7185
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7382394932433974e-05,
+      "loss": 1.1285,
+      "step": 7186
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7358601656754923e-05,
+      "loss": 1.2442,
+      "step": 7187
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7334817084596454e-05,
+      "loss": 1.3896,
+      "step": 7188
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7311041218808375e-05,
+      "loss": 1.328,
+      "step": 7189
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.728727406223934e-05,
+      "loss": 0.7411,
+      "step": 7190
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7263515617737078e-05,
+      "loss": 1.4956,
+      "step": 7191
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7239765888148107e-05,
+      "loss": 1.2275,
+      "step": 7192
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.721602487631807e-05,
+      "loss": 1.366,
+      "step": 7193
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.719229258509144e-05,
+      "loss": 1.5488,
+      "step": 7194
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7168569017311752e-05,
+      "loss": 0.9772,
+      "step": 7195
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.714485417582139e-05,
+      "loss": 0.9899,
+      "step": 7196
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.712114806346181e-05,
+      "loss": 1.3028,
+      "step": 7197
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.709745068307331e-05,
+      "loss": 1.2572,
+      "step": 7198
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.707376203749521e-05,
+      "loss": 1.17,
+      "step": 7199
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.705008212956571e-05,
+      "loss": 1.4161,
+      "step": 7200
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.7026410962122095e-05,
+      "loss": 1.5366,
+      "step": 7201
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.7002748538000457e-05,
+      "loss": 1.4442,
+      "step": 7202
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6979094860035958e-05,
+      "loss": 1.1616,
+      "step": 7203
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6955449931062648e-05,
+      "loss": 1.3788,
+      "step": 7204
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6931813753913514e-05,
+      "loss": 1.4631,
+      "step": 7205
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6908186331420525e-05,
+      "loss": 1.3404,
+      "step": 7206
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.688456766641463e-05,
+      "loss": 1.2912,
+      "step": 7207
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.686095776172566e-05,
+      "loss": 1.0458,
+      "step": 7208
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6837356620182465e-05,
+      "loss": 1.4716,
+      "step": 7209
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6813764244612784e-05,
+      "loss": 1.6728,
+      "step": 7210
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6790180637843367e-05,
+      "loss": 0.999,
+      "step": 7211
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6766605802699872e-05,
+      "loss": 0.9748,
+      "step": 7212
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6743039742006904e-05,
+      "loss": 1.1926,
+      "step": 7213
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6719482458587996e-05,
+      "loss": 1.0938,
+      "step": 7214
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.669593395526573e-05,
+      "loss": 1.5089,
+      "step": 7215
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6672394234861497e-05,
+      "loss": 1.312,
+      "step": 7216
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6648863300195782e-05,
+      "loss": 1.2994,
+      "step": 7217
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6625341154087903e-05,
+      "loss": 1.5909,
+      "step": 7218
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6601827799356116e-05,
+      "loss": 1.1739,
+      "step": 7219
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.657832323881776e-05,
+      "loss": 1.3255,
+      "step": 7220
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6554827475288946e-05,
+      "loss": 1.488,
+      "step": 7221
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.653134051158489e-05,
+      "loss": 1.0386,
+      "step": 7222
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6507862350519618e-05,
+      "loss": 1.0961,
+      "step": 7223
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6484392994906226e-05,
+      "loss": 1.4413,
+      "step": 7224
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.646093244755665e-05,
+      "loss": 1.3981,
+      "step": 7225
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6437480711281837e-05,
+      "loss": 1.3425,
+      "step": 7226
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.64140377888916e-05,
+      "loss": 1.539,
+      "step": 7227
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6390603683194816e-05,
+      "loss": 1.211,
+      "step": 7228
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.636717839699917e-05,
+      "loss": 1.2181,
+      "step": 7229
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6343761933111444e-05,
+      "loss": 1.1033,
+      "step": 7230
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6320354294337234e-05,
+      "loss": 1.2742,
+      "step": 7231
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6296955483481134e-05,
+      "loss": 1.2333,
+      "step": 7232
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.627356550334662e-05,
+      "loss": 1.3811,
+      "step": 7233
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6250184356736228e-05,
+      "loss": 0.8385,
+      "step": 7234
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.622681204645132e-05,
+      "loss": 1.3319,
+      "step": 7235
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.62034485752923e-05,
+      "loss": 1.4674,
+      "step": 7236
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.618009394605839e-05,
+      "loss": 1.597,
+      "step": 7237
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6156748161547927e-05,
+      "loss": 1.1784,
+      "step": 7238
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6133411224557948e-05,
+      "loss": 1.0686,
+      "step": 7239
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6110083137884668e-05,
+      "loss": 1.2189,
+      "step": 7240
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6086763904323065e-05,
+      "loss": 1.2672,
+      "step": 7241
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6063453526667204e-05,
+      "loss": 1.4104,
+      "step": 7242
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.6040152007709952e-05,
+      "loss": 1.0879,
+      "step": 7243
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.6016859350243227e-05,
+      "loss": 1.5939,
+      "step": 7244
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5993575557057824e-05,
+      "loss": 1.6495,
+      "step": 7245
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.597030063094348e-05,
+      "loss": 1.1706,
+      "step": 7246
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5947034574688843e-05,
+      "loss": 1.4503,
+      "step": 7247
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5923777391081594e-05,
+      "loss": 1.1361,
+      "step": 7248
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5900529082908242e-05,
+      "loss": 1.3622,
+      "step": 7249
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5877289652954318e-05,
+      "loss": 1.2935,
+      "step": 7250
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5854059104004236e-05,
+      "loss": 1.6446,
+      "step": 7251
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5830837438841337e-05,
+      "loss": 1.2696,
+      "step": 7252
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5807624660247965e-05,
+      "loss": 1.3669,
+      "step": 7253
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5784420771005314e-05,
+      "loss": 1.2848,
+      "step": 7254
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5761225773893606e-05,
+      "loss": 1.1633,
+      "step": 7255
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5738039671691882e-05,
+      "loss": 1.3183,
+      "step": 7256
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5714862467178247e-05,
+      "loss": 1.3849,
+      "step": 7257
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5691694163129652e-05,
+      "loss": 1.3868,
+      "step": 7258
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5668534762321995e-05,
+      "loss": 1.1825,
+      "step": 7259
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.564538426753009e-05,
+      "loss": 1.1221,
+      "step": 7260
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5622242681527776e-05,
+      "loss": 1.044,
+      "step": 7261
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.559911000708769e-05,
+      "loss": 1.252,
+      "step": 7262
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.557598624698154e-05,
+      "loss": 1.0355,
+      "step": 7263
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.555287140397986e-05,
+      "loss": 1.1327,
+      "step": 7264
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5529765480852152e-05,
+      "loss": 1.2117,
+      "step": 7265
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5506668480366825e-05,
+      "loss": 1.4454,
+      "step": 7266
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5483580405291306e-05,
+      "loss": 1.3872,
+      "step": 7267
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5460501258391813e-05,
+      "loss": 1.181,
+      "step": 7268
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5437431042433646e-05,
+      "loss": 1.1419,
+      "step": 7269
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.54143697601809e-05,
+      "loss": 1.2916,
+      "step": 7270
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.539131741439671e-05,
+      "loss": 1.0816,
+      "step": 7271
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5368274007843072e-05,
+      "loss": 1.0679,
+      "step": 7272
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5345239543280918e-05,
+      "loss": 1.4106,
+      "step": 7273
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5322214023470102e-05,
+      "loss": 1.0541,
+      "step": 7274
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5299197451169475e-05,
+      "loss": 0.9744,
+      "step": 7275
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.527618982913671e-05,
+      "loss": 1.0865,
+      "step": 7276
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.525319116012851e-05,
+      "loss": 1.3736,
+      "step": 7277
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.523020144690045e-05,
+      "loss": 1.1665,
+      "step": 7278
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5207220692207023e-05,
+      "loss": 1.373,
+      "step": 7279
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5184248898801632e-05,
+      "loss": 1.4305,
+      "step": 7280
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.516128606943672e-05,
+      "loss": 1.3896,
+      "step": 7281
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.513833220686349e-05,
+      "loss": 1.2821,
+      "step": 7282
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.511538731383224e-05,
+      "loss": 1.2167,
+      "step": 7283
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5092451393092033e-05,
+      "loss": 1.1477,
+      "step": 7284
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5069524447391002e-05,
+      "loss": 1.35,
+      "step": 7285
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5046606479476096e-05,
+      "loss": 1.4341,
+      "step": 7286
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5023697492093244e-05,
+      "loss": 0.9331,
+      "step": 7287
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.500079748798725e-05,
+      "loss": 1.4182,
+      "step": 7288
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4977906469901902e-05,
+      "loss": 1.0272,
+      "step": 7289
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.495502444057991e-05,
+      "loss": 1.6786,
+      "step": 7290
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.493215140276286e-05,
+      "loss": 1.5931,
+      "step": 7291
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.490928735919127e-05,
+      "loss": 1.1146,
+      "step": 7292
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.488643231260458e-05,
+      "loss": 1.212,
+      "step": 7293
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4863586265741214e-05,
+      "loss": 1.4658,
+      "step": 7294
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.484074922133841e-05,
+      "loss": 1.368,
+      "step": 7295
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4817921182132442e-05,
+      "loss": 1.2698,
+      "step": 7296
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4795102150858395e-05,
+      "loss": 1.2366,
+      "step": 7297
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.477229213025042e-05,
+      "loss": 1.4563,
+      "step": 7298
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4749491123041367e-05,
+      "loss": 0.7145,
+      "step": 7299
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4726699131963226e-05,
+      "loss": 1.6306,
+      "step": 7300
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4703916159746775e-05,
+      "loss": 1.3882,
+      "step": 7301
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.46811422091218e-05,
+      "loss": 1.1913,
+      "step": 7302
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4658377282816903e-05,
+      "loss": 1.4747,
+      "step": 7303
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4635621383559716e-05,
+      "loss": 1.1089,
+      "step": 7304
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.46128745140767e-05,
+      "loss": 1.468,
+      "step": 7305
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4590136677093278e-05,
+      "loss": 1.5165,
+      "step": 7306
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.456740787533376e-05,
+      "loss": 1.38,
+      "step": 7307
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.454468811152143e-05,
+      "loss": 1.16,
+      "step": 7308
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4521977388378413e-05,
+      "loss": 1.2465,
+      "step": 7309
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4499275708625835e-05,
+      "loss": 1.2329,
+      "step": 7310
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4476583074983696e-05,
+      "loss": 1.1687,
+      "step": 7311
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4453899490170872e-05,
+      "loss": 1.1926,
+      "step": 7312
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4431224956905196e-05,
+      "loss": 1.2801,
+      "step": 7313
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4408559477903446e-05,
+      "loss": 1.2071,
+      "step": 7314
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4385903055881255e-05,
+      "loss": 1.5043,
+      "step": 7315
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4363255693553245e-05,
+      "loss": 1.56,
+      "step": 7316
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4340617393632835e-05,
+      "loss": 1.4568,
+      "step": 7317
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4317988158832505e-05,
+      "loss": 1.4608,
+      "step": 7318
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4295367991863537e-05,
+      "loss": 0.9963,
+      "step": 7319
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4272756895436176e-05,
+      "loss": 1.1975,
+      "step": 7320
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4250154872259524e-05,
+      "loss": 1.1511,
+      "step": 7321
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.422756192504171e-05,
+      "loss": 1.2237,
+      "step": 7322
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4204978056489636e-05,
+      "loss": 1.2681,
+      "step": 7323
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.418240326930924e-05,
+      "loss": 1.2652,
+      "step": 7324
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4159837566205303e-05,
+      "loss": 1.4141,
+      "step": 7325
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4137280949881502e-05,
+      "loss": 1.6877,
+      "step": 7326
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4114733423040504e-05,
+      "loss": 1.0393,
+      "step": 7327
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4092194988383786e-05,
+      "loss": 0.9731,
+      "step": 7328
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.406966564861185e-05,
+      "loss": 1.0615,
+      "step": 7329
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4047145406423978e-05,
+      "loss": 1.2342,
+      "step": 7330
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4024634264518486e-05,
+      "loss": 1.5296,
+      "step": 7331
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4002132225592534e-05,
+      "loss": 1.2386,
+      "step": 7332
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.397963929234217e-05,
+      "loss": 1.3247,
+      "step": 7333
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3957155467462388e-05,
+      "loss": 1.445,
+      "step": 7334
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.393468075364712e-05,
+      "loss": 1.1571,
+      "step": 7335
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3912215153589136e-05,
+      "loss": 0.998,
+      "step": 7336
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3889758669980166e-05,
+      "loss": 1.0246,
+      "step": 7337
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3867311305510843e-05,
+      "loss": 1.1642,
+      "step": 7338
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3844873062870677e-05,
+      "loss": 1.2235,
+      "step": 7339
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3822443944748086e-05,
+      "loss": 1.0753,
+      "step": 7340
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.380002395383044e-05,
+      "loss": 1.3503,
+      "step": 7341
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3777613092803962e-05,
+      "loss": 1.2238,
+      "step": 7342
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3755211364353857e-05,
+      "loss": 1.2669,
+      "step": 7343
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.373281877116412e-05,
+      "loss": 1.2988,
+      "step": 7344
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.371043531591779e-05,
+      "loss": 1.255,
+      "step": 7345
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.368806100129668e-05,
+      "loss": 1.1745,
+      "step": 7346
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3665695829981594e-05,
+      "loss": 1.1859,
+      "step": 7347
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3643339804652166e-05,
+      "loss": 1.4134,
+      "step": 7348
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.362099292798705e-05,
+      "loss": 1.1677,
+      "step": 7349
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3598655202663676e-05,
+      "loss": 1.3707,
+      "step": 7350
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3576326631358492e-05,
+      "loss": 1.0911,
+      "step": 7351
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3554007216746754e-05,
+      "loss": 1.2819,
+      "step": 7352
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.353169696150267e-05,
+      "loss": 1.3036,
+      "step": 7353
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.350939586829931e-05,
+      "loss": 1.5102,
+      "step": 7354
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.348710393980873e-05,
+      "loss": 1.2666,
+      "step": 7355
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.346482117870179e-05,
+      "loss": 1.0554,
+      "step": 7356
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3442547587648345e-05,
+      "loss": 1.2203,
+      "step": 7357
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.342028316931707e-05,
+      "loss": 1.0582,
+      "step": 7358
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3398027926375576e-05,
+      "loss": 1.3541,
+      "step": 7359
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3375781861490355e-05,
+      "loss": 1.3285,
+      "step": 7360
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3353544977326835e-05,
+      "loss": 1.4822,
+      "step": 7361
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3331317276549347e-05,
+      "loss": 1.396,
+      "step": 7362
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3309098761821058e-05,
+      "loss": 1.1635,
+      "step": 7363
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3286889435804125e-05,
+      "loss": 1.3412,
+      "step": 7364
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3264689301159536e-05,
+      "loss": 1.0255,
+      "step": 7365
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.324249836054718e-05,
+      "loss": 1.5121,
+      "step": 7366
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.322031661662586e-05,
+      "loss": 1.1454,
+      "step": 7367
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3198144072053306e-05,
+      "loss": 1.2488,
+      "step": 7368
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.317598072948608e-05,
+      "loss": 1.2323,
+      "step": 7369
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3153826591579743e-05,
+      "loss": 1.108,
+      "step": 7370
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3131681660988647e-05,
+      "loss": 1.1157,
+      "step": 7371
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3109545940366084e-05,
+      "loss": 1.3071,
+      "step": 7372
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.308741943236423e-05,
+      "loss": 1.2608,
+      "step": 7373
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3065302139634215e-05,
+      "loss": 1.3777,
+      "step": 7374
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3043194064825956e-05,
+      "loss": 1.3694,
+      "step": 7375
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3021095210588405e-05,
+      "loss": 1.274,
+      "step": 7376
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2999005579569265e-05,
+      "loss": 1.3814,
+      "step": 7377
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2976925174415263e-05,
+      "loss": 1.3254,
+      "step": 7378
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.295485399777193e-05,
+      "loss": 1.3469,
+      "step": 7379
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2932792052283725e-05,
+      "loss": 1.1787,
+      "step": 7380
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2910739340593958e-05,
+      "loss": 1.4668,
+      "step": 7381
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2888695865344945e-05,
+      "loss": 1.0164,
+      "step": 7382
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2866661629177756e-05,
+      "loss": 1.3913,
+      "step": 7383
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2844636634732474e-05,
+      "loss": 1.2068,
+      "step": 7384
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2822620884647995e-05,
+      "loss": 1.1468,
+      "step": 7385
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2800614381562137e-05,
+      "loss": 0.7653,
+      "step": 7386
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2778617128111577e-05,
+      "loss": 1.4722,
+      "step": 7387
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.275662912693197e-05,
+      "loss": 1.0593,
+      "step": 7388
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2734650380657763e-05,
+      "loss": 1.0221,
+      "step": 7389
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2712680891922365e-05,
+      "loss": 1.121,
+      "step": 7390
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.269072066335801e-05,
+      "loss": 1.2183,
+      "step": 7391
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2668769697595916e-05,
+      "loss": 1.183,
+      "step": 7392
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2646827997266094e-05,
+      "loss": 0.9746,
+      "step": 7393
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2624895564997505e-05,
+      "loss": 1.3638,
+      "step": 7394
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2602972403417934e-05,
+      "loss": 0.3007,
+      "step": 7395
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.258105851515415e-05,
+      "loss": 0.8752,
+      "step": 7396
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2559153902831775e-05,
+      "loss": 1.2484,
+      "step": 7397
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.253725856907528e-05,
+      "loss": 0.9729,
+      "step": 7398
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2515372516508072e-05,
+      "loss": 1.5466,
+      "step": 7399
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.249349574775237e-05,
+      "loss": 1.524,
+      "step": 7400
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2471628265429412e-05,
+      "loss": 1.2738,
+      "step": 7401
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.244977007215918e-05,
+      "loss": 1.2038,
+      "step": 7402
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2427921170560685e-05,
+      "loss": 0.9918,
+      "step": 7403
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.240608156325167e-05,
+      "loss": 1.1427,
+      "step": 7404
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2384251252848907e-05,
+      "loss": 1.2489,
+      "step": 7405
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.236243024196798e-05,
+      "loss": 1.1651,
+      "step": 7406
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2340618533223358e-05,
+      "loss": 1.2848,
+      "step": 7407
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2318816129228382e-05,
+      "loss": 1.3545,
+      "step": 7408
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2297023032595366e-05,
+      "loss": 1.3587,
+      "step": 7409
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2275239245935386e-05,
+      "loss": 1.2808,
+      "step": 7410
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2253464771858513e-05,
+      "loss": 1.3703,
+      "step": 7411
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2231699612973634e-05,
+      "loss": 1.517,
+      "step": 7412
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.220994377188854e-05,
+      "loss": 1.2383,
+      "step": 7413
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2188197251209874e-05,
+      "loss": 1.4919,
+      "step": 7414
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.216646005354326e-05,
+      "loss": 1.0772,
+      "step": 7415
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2144732181493054e-05,
+      "loss": 1.111,
+      "step": 7416
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.212301363766266e-05,
+      "loss": 1.1581,
+      "step": 7417
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2101304424654244e-05,
+      "loss": 1.0729,
+      "step": 7418
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.207960454506889e-05,
+      "loss": 1.2452,
+      "step": 7419
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2057914001506552e-05,
+      "loss": 1.2085,
+      "step": 7420
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2036232796566126e-05,
+      "loss": 1.4961,
+      "step": 7421
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2014560932845284e-05,
+      "loss": 1.0247,
+      "step": 7422
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.19928984129407e-05,
+      "loss": 1.247,
+      "step": 7423
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1971245239447802e-05,
+      "loss": 1.0795,
+      "step": 7424
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.194960141496103e-05,
+      "loss": 1.3548,
+      "step": 7425
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.192796694207361e-05,
+      "loss": 1.1861,
+      "step": 7426
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.190634182337765e-05,
+      "loss": 1.2026,
+      "step": 7427
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.188472606146417e-05,
+      "loss": 1.2134,
+      "step": 7428
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.186311965892308e-05,
+      "loss": 1.3313,
+      "step": 7429
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1841522618343112e-05,
+      "loss": 1.2147,
+      "step": 7430
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.181993494231197e-05,
+      "loss": 1.2361,
+      "step": 7431
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1798356633416138e-05,
+      "loss": 1.3771,
+      "step": 7432
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1776787694241006e-05,
+      "loss": 1.0269,
+      "step": 7433
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1755228127370907e-05,
+      "loss": 1.1683,
+      "step": 7434
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.173367793538893e-05,
+      "loss": 1.2829,
+      "step": 7435
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1712137120877175e-05,
+      "loss": 0.8411,
+      "step": 7436
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.16906056864165e-05,
+      "loss": 1.7318,
+      "step": 7437
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.166908363458673e-05,
+      "loss": 1.5301,
+      "step": 7438
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.164757096796651e-05,
+      "loss": 1.0144,
+      "step": 7439
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1626067689133388e-05,
+      "loss": 1.3376,
+      "step": 7440
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.160457380066373e-05,
+      "loss": 0.9962,
+      "step": 7441
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1583089305132886e-05,
+      "loss": 1.1612,
+      "step": 7442
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1561614205114966e-05,
+      "loss": 1.4838,
+      "step": 7443
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1540148503183065e-05,
+      "loss": 1.4999,
+      "step": 7444
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1518692201909062e-05,
+      "loss": 1.1227,
+      "step": 7445
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1497245303863744e-05,
+      "loss": 1.3266,
+      "step": 7446
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1475807811616732e-05,
+      "loss": 1.1223,
+      "step": 7447
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.145437972773662e-05,
+      "loss": 1.5672,
+      "step": 7448
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1432961054790746e-05,
+      "loss": 1.2862,
+      "step": 7449
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.141155179534545e-05,
+      "loss": 1.4055,
+      "step": 7450
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.139015195196583e-05,
+      "loss": 1.2962,
+      "step": 7451
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1368761527215964e-05,
+      "loss": 1.3098,
+      "step": 7452
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.134738052365869e-05,
+      "loss": 1.1822,
+      "step": 7453
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.132600894385579e-05,
+      "loss": 1.5558,
+      "step": 7454
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1304646790367865e-05,
+      "loss": 1.2834,
+      "step": 7455
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1283294065754477e-05,
+      "loss": 1.4637,
+      "step": 7456
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1261950772573947e-05,
+      "loss": 1.3899,
+      "step": 7457
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.124061691338357e-05,
+      "loss": 1.0551,
+      "step": 7458
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1219292490739428e-05,
+      "loss": 1.4476,
+      "step": 7459
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1197977507196508e-05,
+      "loss": 1.7531,
+      "step": 7460
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1176671965308648e-05,
+      "loss": 1.5279,
+      "step": 7461
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1155375867628603e-05,
+      "loss": 1.1461,
+      "step": 7462
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.113408921670791e-05,
+      "loss": 1.324,
+      "step": 7463
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1112812015097093e-05,
+      "loss": 1.146,
+      "step": 7464
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.109154426534542e-05,
+      "loss": 1.4309,
+      "step": 7465
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1070285970001158e-05,
+      "loss": 1.5755,
+      "step": 7466
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.104903713161127e-05,
+      "loss": 1.1868,
+      "step": 7467
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1027797752721735e-05,
+      "loss": 1.3201,
+      "step": 7468
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1006567835877357e-05,
+      "loss": 1.183,
+      "step": 7469
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0985347383621766e-05,
+      "loss": 1.7771,
+      "step": 7470
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0964136398497526e-05,
+      "loss": 1.585,
+      "step": 7471
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0942934883046005e-05,
+      "loss": 1.2516,
+      "step": 7472
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0921742839807456e-05,
+      "loss": 1.0641,
+      "step": 7473
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.090056027132099e-05,
+      "loss": 1.0843,
+      "step": 7474
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0879387180124632e-05,
+      "loss": 1.3334,
+      "step": 7475
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0858223568755186e-05,
+      "loss": 0.9399,
+      "step": 7476
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0837069439748426e-05,
+      "loss": 1.1546,
+      "step": 7477
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0815924795638885e-05,
+      "loss": 1.4426,
+      "step": 7478
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.079478963896003e-05,
+      "loss": 1.3729,
+      "step": 7479
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.077366397224413e-05,
+      "loss": 1.2638,
+      "step": 7480
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0752547798022405e-05,
+      "loss": 1.1537,
+      "step": 7481
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0731441118824846e-05,
+      "loss": 1.4204,
+      "step": 7482
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.071034393718039e-05,
+      "loss": 1.1519,
+      "step": 7483
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.068925625561673e-05,
+      "loss": 1.3711,
+      "step": 7484
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.066817807666056e-05,
+      "loss": 1.4607,
+      "step": 7485
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.064710940283732e-05,
+      "loss": 1.6137,
+      "step": 7486
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0626050236671345e-05,
+      "loss": 1.2401,
+      "step": 7487
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.060500058068582e-05,
+      "loss": 1.426,
+      "step": 7488
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.058396043740286e-05,
+      "loss": 0.961,
+      "step": 7489
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0562929809343334e-05,
+      "loss": 1.5743,
+      "step": 7490
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0541908699027078e-05,
+      "loss": 1.1522,
+      "step": 7491
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0520897108972703e-05,
+      "loss": 1.4775,
+      "step": 7492
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0499895041697704e-05,
+      "loss": 1.3133,
+      "step": 7493
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0478902499718434e-05,
+      "loss": 1.4761,
+      "step": 7494
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0457919485550136e-05,
+      "loss": 1.3053,
+      "step": 7495
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.043694600170687e-05,
+      "loss": 1.6724,
+      "step": 7496
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.04159820507016e-05,
+      "loss": 1.0467,
+      "step": 7497
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.039502763504607e-05,
+      "loss": 1.2703,
+      "step": 7498
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0374082757250988e-05,
+      "loss": 0.9237,
+      "step": 7499
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0353147419825834e-05,
+      "loss": 1.3207,
+      "step": 7500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.033222162527898e-05,
+      "loss": 0.9789,
+      "step": 7501
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0311305376117608e-05,
+      "loss": 1.4073,
+      "step": 7502
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0290398674847855e-05,
+      "loss": 1.3401,
+      "step": 7503
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.02695015239746e-05,
+      "loss": 1.1275,
+      "step": 7504
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.024861392600169e-05,
+      "loss": 1.3793,
+      "step": 7505
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.022773588343174e-05,
+      "loss": 1.3698,
+      "step": 7506
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0206867398766226e-05,
+      "loss": 1.2903,
+      "step": 7507
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0186008474505558e-05,
+      "loss": 1.2534,
+      "step": 7508
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.016515911314888e-05,
+      "loss": 1.228,
+      "step": 7509
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0144319317194328e-05,
+      "loss": 1.1783,
+      "step": 7510
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0123489089138757e-05,
+      "loss": 1.4237,
+      "step": 7511
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.010266843147799e-05,
+      "loss": 1.6231,
+      "step": 7512
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0081857346706622e-05,
+      "loss": 1.3102,
+      "step": 7513
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0061055837318144e-05,
+      "loss": 1.3382,
+      "step": 7514
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0040263905804847e-05,
+      "loss": 1.4031,
+      "step": 7515
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0019481554657972e-05,
+      "loss": 1.0679,
+      "step": 7516
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9998708786367493e-05,
+      "loss": 1.4784,
+      "step": 7517
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9977945603422365e-05,
+      "loss": 1.2667,
+      "step": 7518
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.995719200831029e-05,
+      "loss": 1.3293,
+      "step": 7519
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.993644800351786e-05,
+      "loss": 1.2655,
+      "step": 7520
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9915713591530482e-05,
+      "loss": 1.4068,
+      "step": 7521
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.989498877483251e-05,
+      "loss": 1.1735,
+      "step": 7522
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9874273555907032e-05,
+      "loss": 1.1222,
+      "step": 7523
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9853567937236094e-05,
+      "loss": 1.1528,
+      "step": 7524
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9832871921300466e-05,
+      "loss": 1.1758,
+      "step": 7525
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.981218551057994e-05,
+      "loss": 1.4662,
+      "step": 7526
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.979150870755294e-05,
+      "loss": 1.1777,
+      "step": 7527
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9770841514696937e-05,
+      "loss": 0.9891,
+      "step": 7528
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.975018393448812e-05,
+      "loss": 1.5848,
+      "step": 7529
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9729535969401612e-05,
+      "loss": 1.2104,
+      "step": 7530
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9708897621911304e-05,
+      "loss": 1.1496,
+      "step": 7531
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9688268894490037e-05,
+      "loss": 1.7061,
+      "step": 7532
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.96676497896094e-05,
+      "loss": 0.9905,
+      "step": 7533
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9647040309739874e-05,
+      "loss": 0.9815,
+      "step": 7534
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.962644045735076e-05,
+      "loss": 1.0028,
+      "step": 7535
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.960585023491027e-05,
+      "loss": 1.2111,
+      "step": 7536
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9585269644885372e-05,
+      "loss": 1.9513,
+      "step": 7537
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.956469868974198e-05,
+      "loss": 1.1585,
+      "step": 7538
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.954413737194477e-05,
+      "loss": 1.1982,
+      "step": 7539
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9523585693957268e-05,
+      "loss": 1.4467,
+      "step": 7540
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9503043658241938e-05,
+      "loss": 1.5039,
+      "step": 7541
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.948251126725995e-05,
+      "loss": 1.1714,
+      "step": 7542
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9461988523471452e-05,
+      "loss": 1.1995,
+      "step": 7543
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.944147542933532e-05,
+      "loss": 1.3859,
+      "step": 7544
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.942097198730938e-05,
+      "loss": 1.0814,
+      "step": 7545
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9400478199850215e-05,
+      "loss": 1.2781,
+      "step": 7546
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9379994069413297e-05,
+      "loss": 1.2569,
+      "step": 7547
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9359519598452903e-05,
+      "loss": 1.4907,
+      "step": 7548
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.933905478942223e-05,
+      "loss": 0.9845,
+      "step": 7549
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9318599644773215e-05,
+      "loss": 1.3678,
+      "step": 7550
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9298154166956738e-05,
+      "loss": 1.4692,
+      "step": 7551
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.927771835842245e-05,
+      "loss": 1.2963,
+      "step": 7552
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.925729222161885e-05,
+      "loss": 1.5808,
+      "step": 7553
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.923687575899329e-05,
+      "loss": 1.2026,
+      "step": 7554
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9216468972991995e-05,
+      "loss": 1.6025,
+      "step": 7555
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9196071866059972e-05,
+      "loss": 1.1809,
+      "step": 7556
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9175684440641138e-05,
+      "loss": 1.186,
+      "step": 7557
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9155306699178165e-05,
+      "loss": 1.1653,
+      "step": 7558
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9134938644112655e-05,
+      "loss": 1.1512,
+      "step": 7559
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9114580277884974e-05,
+      "loss": 0.9633,
+      "step": 7560
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9094231602934365e-05,
+      "loss": 1.2743,
+      "step": 7561
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9073892621698875e-05,
+      "loss": 1.6587,
+      "step": 7562
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9053563336615466e-05,
+      "loss": 1.0183,
+      "step": 7563
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9033243750119832e-05,
+      "loss": 1.08,
+      "step": 7564
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.901293386464663e-05,
+      "loss": 1.3863,
+      "step": 7565
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8992633682629245e-05,
+      "loss": 1.2084,
+      "step": 7566
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8972343206499953e-05,
+      "loss": 0.9954,
+      "step": 7567
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8952062438689822e-05,
+      "loss": 1.128,
+      "step": 7568
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8931791381628837e-05,
+      "loss": 1.1593,
+      "step": 7569
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.891153003774573e-05,
+      "loss": 1.5496,
+      "step": 7570
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.889127840946816e-05,
+      "loss": 1.4618,
+      "step": 7571
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8871036499222518e-05,
+      "loss": 1.5134,
+      "step": 7572
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8850804309434166e-05,
+      "loss": 1.1855,
+      "step": 7573
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.883058184252712e-05,
+      "loss": 1.416,
+      "step": 7574
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8810369100924407e-05,
+      "loss": 1.1445,
+      "step": 7575
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.879016608704778e-05,
+      "loss": 1.1339,
+      "step": 7576
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.876997280331785e-05,
+      "loss": 1.5009,
+      "step": 7577
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8749789252154127e-05,
+      "loss": 1.4204,
+      "step": 7578
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8729615435974867e-05,
+      "loss": 1.5641,
+      "step": 7579
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8709451357197206e-05,
+      "loss": 1.2693,
+      "step": 7580
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8689297018237052e-05,
+      "loss": 1.32,
+      "step": 7581
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.866915242150925e-05,
+      "loss": 0.9243,
+      "step": 7582
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8649017569427396e-05,
+      "loss": 1.1467,
+      "step": 7583
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.862889246440397e-05,
+      "loss": 1.0003,
+      "step": 7584
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8608777108850206e-05,
+      "loss": 1.4026,
+      "step": 7585
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8588671505176326e-05,
+      "loss": 1.428,
+      "step": 7586
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8568575655791155e-05,
+      "loss": 1.4255,
+      "step": 7587
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8548489563102567e-05,
+      "loss": 1.7193,
+      "step": 7588
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8528413229517105e-05,
+      "loss": 1.4389,
+      "step": 7589
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8508346657440278e-05,
+      "loss": 1.159,
+      "step": 7590
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8488289849276308e-05,
+      "loss": 1.6168,
+      "step": 7591
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.846824280742835e-05,
+      "loss": 1.1832,
+      "step": 7592
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.84482055342983e-05,
+      "loss": 1.5018,
+      "step": 7593
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.842817803228695e-05,
+      "loss": 1.5925,
+      "step": 7594
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.840816030379383e-05,
+      "loss": 1.3106,
+      "step": 7595
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8388152351217434e-05,
+      "loss": 1.5101,
+      "step": 7596
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8368154176954967e-05,
+      "loss": 1.1553,
+      "step": 7597
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8348165783402547e-05,
+      "loss": 1.5415,
+      "step": 7598
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8328187172955057e-05,
+      "loss": 1.2713,
+      "step": 7599
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8308218348006246e-05,
+      "loss": 1.1826,
+      "step": 7600
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.828825931094863e-05,
+      "loss": 1.1566,
+      "step": 7601
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8268310064173655e-05,
+      "loss": 1.4031,
+      "step": 7602
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8248370610071497e-05,
+      "loss": 1.266,
+      "step": 7603
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8228440951031245e-05,
+      "loss": 1.4278,
+      "step": 7604
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.820852108944071e-05,
+      "loss": 1.4722,
+      "step": 7605
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.818861102768665e-05,
+      "loss": 1.1365,
+      "step": 7606
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.816871076815455e-05,
+      "loss": 1.4179,
+      "step": 7607
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8148820313228775e-05,
+      "loss": 1.3023,
+      "step": 7608
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8128939665292455e-05,
+      "loss": 0.9602,
+      "step": 7609
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8109068826727648e-05,
+      "loss": 1.4969,
+      "step": 7610
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.808920779991512e-05,
+      "loss": 1.4217,
+      "step": 7611
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8069356587234566e-05,
+      "loss": 1.4263,
+      "step": 7612
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8049515191064448e-05,
+      "loss": 1.0668,
+      "step": 7613
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.802968361378201e-05,
+      "loss": 1.217,
+      "step": 7614
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8009861857763444e-05,
+      "loss": 1.4957,
+      "step": 7615
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7990049925383623e-05,
+      "loss": 1.4616,
+      "step": 7616
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7970247819016372e-05,
+      "loss": 1.3796,
+      "step": 7617
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.795045554103422e-05,
+      "loss": 1.0942,
+      "step": 7618
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7930673093808635e-05,
+      "loss": 1.1024,
+      "step": 7619
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7910900479709823e-05,
+      "loss": 1.3844,
+      "step": 7620
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.789113770110683e-05,
+      "loss": 1.1473,
+      "step": 7621
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7871384760367517e-05,
+      "loss": 1.3534,
+      "step": 7622
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.785164165985862e-05,
+      "loss": 1.5659,
+      "step": 7623
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7831908401945607e-05,
+      "loss": 1.1564,
+      "step": 7624
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7812184988992865e-05,
+      "loss": 1.1705,
+      "step": 7625
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7792471423363544e-05,
+      "loss": 1.3929,
+      "step": 7626
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7772767707419603e-05,
+      "loss": 1.2693,
+      "step": 7627
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7753073843521817e-05,
+      "loss": 1.2651,
+      "step": 7628
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.773338983402987e-05,
+      "loss": 1.2473,
+      "step": 7629
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7713715681302136e-05,
+      "loss": 1.0476,
+      "step": 7630
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.769405138769592e-05,
+      "loss": 1.3929,
+      "step": 7631
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.767439695556725e-05,
+      "loss": 1.4816,
+      "step": 7632
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.76547523872711e-05,
+      "loss": 1.5802,
+      "step": 7633
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7635117685161085e-05,
+      "loss": 1.1981,
+      "step": 7634
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7615492851589787e-05,
+      "loss": 1.2671,
+      "step": 7635
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.759587788890853e-05,
+      "loss": 1.0609,
+      "step": 7636
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7576272799467507e-05,
+      "loss": 1.4604,
+      "step": 7637
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7556677585615667e-05,
+      "loss": 1.1758,
+      "step": 7638
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7537092249700837e-05,
+      "loss": 1.5203,
+      "step": 7639
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7517516794069634e-05,
+      "loss": 0.9909,
+      "step": 7640
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7497951221067464e-05,
+      "loss": 1.1935,
+      "step": 7641
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.747839553303857e-05,
+      "loss": 1.2616,
+      "step": 7642
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7458849732326045e-05,
+      "loss": 1.4711,
+      "step": 7643
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.743931382127173e-05,
+      "loss": 1.5659,
+      "step": 7644
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.741978780221636e-05,
+      "loss": 0.9215,
+      "step": 7645
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7400271677499426e-05,
+      "loss": 1.0151,
+      "step": 7646
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7380765449459245e-05,
+      "loss": 1.4975,
+      "step": 7647
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.736126912043292e-05,
+      "loss": 1.1065,
+      "step": 7648
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7341782692756448e-05,
+      "loss": 1.4417,
+      "step": 7649
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.73223061687646e-05,
+      "loss": 1.2078,
+      "step": 7650
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7302839550790905e-05,
+      "loss": 1.0519,
+      "step": 7651
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7283382841167815e-05,
+      "loss": 1.4368,
+      "step": 7652
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.726393604222649e-05,
+      "loss": 1.1744,
+      "step": 7653
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7244499156296967e-05,
+      "loss": 1.2307,
+      "step": 7654
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7225072185708036e-05,
+      "loss": 1.1971,
+      "step": 7655
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.720565513278739e-05,
+      "loss": 1.1621,
+      "step": 7656
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.718624799986143e-05,
+      "loss": 1.2386,
+      "step": 7657
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7166850789255474e-05,
+      "loss": 1.2617,
+      "step": 7658
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.714746350329356e-05,
+      "loss": 1.2741,
+      "step": 7659
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7128086144298593e-05,
+      "loss": 1.4131,
+      "step": 7660
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.710871871459222e-05,
+      "loss": 0.9455,
+      "step": 7661
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7089361216495014e-05,
+      "loss": 1.3733,
+      "step": 7662
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7070013652326246e-05,
+      "loss": 1.314,
+      "step": 7663
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7050676024404066e-05,
+      "loss": 1.3231,
+      "step": 7664
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.7031348335045395e-05,
+      "loss": 1.5103,
+      "step": 7665
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.7012030586565997e-05,
+      "loss": 0.8623,
+      "step": 7666
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6992722781280413e-05,
+      "loss": 1.1258,
+      "step": 7667
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6973424921502012e-05,
+      "loss": 1.1436,
+      "step": 7668
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6954137009542926e-05,
+      "loss": 1.1522,
+      "step": 7669
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6934859047714203e-05,
+      "loss": 1.3008,
+      "step": 7670
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.691559103832556e-05,
+      "loss": 1.3583,
+      "step": 7671
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6896332983685647e-05,
+      "loss": 1.3583,
+      "step": 7672
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6877084886101845e-05,
+      "loss": 1.2829,
+      "step": 7673
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6857846747880367e-05,
+      "loss": 1.1883,
+      "step": 7674
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6838618571326194e-05,
+      "loss": 1.5702,
+      "step": 7675
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6819400358743197e-05,
+      "loss": 1.1326,
+      "step": 7676
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6800192112433954e-05,
+      "loss": 1.2239,
+      "step": 7677
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6780993834699953e-05,
+      "loss": 1.1317,
+      "step": 7678
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.676180552784139e-05,
+      "loss": 1.4147,
+      "step": 7679
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.674262719415738e-05,
+      "loss": 1.6733,
+      "step": 7680
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.672345883594566e-05,
+      "loss": 1.1261,
+      "step": 7681
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6704300455502974e-05,
+      "loss": 1.3904,
+      "step": 7682
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.668515205512473e-05,
+      "loss": 1.4751,
+      "step": 7683
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6666013637105204e-05,
+      "loss": 1.4032,
+      "step": 7684
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6646885203737506e-05,
+      "loss": 1.2526,
+      "step": 7685
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6627766757313466e-05,
+      "loss": 1.3923,
+      "step": 7686
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.660865830012378e-05,
+      "loss": 1.179,
+      "step": 7687
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6589559834457868e-05,
+      "loss": 1.2343,
+      "step": 7688
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.657047136260409e-05,
+      "loss": 0.9599,
+      "step": 7689
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.655139288684947e-05,
+      "loss": 1.5624,
+      "step": 7690
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6532324409479938e-05,
+      "loss": 1.2696,
+      "step": 7691
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.651326593278013e-05,
+      "loss": 1.0891,
+      "step": 7692
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6494217459033624e-05,
+      "loss": 1.3034,
+      "step": 7693
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.647517899052259e-05,
+      "loss": 1.2141,
+      "step": 7694
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6456150529528214e-05,
+      "loss": 1.3709,
+      "step": 7695
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.643713207833033e-05,
+      "loss": 1.205,
+      "step": 7696
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6418123639207682e-05,
+      "loss": 1.3924,
+      "step": 7697
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6399125214437716e-05,
+      "loss": 1.3493,
+      "step": 7698
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6380136806296776e-05,
+      "loss": 1.151,
+      "step": 7699
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6361158417059918e-05,
+      "loss": 1.0384,
+      "step": 7700
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.634219004900106e-05,
+      "loss": 1.429,
+      "step": 7701
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.632323170439285e-05,
+      "loss": 1.431,
+      "step": 7702
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6304283385506835e-05,
+      "loss": 1.1335,
+      "step": 7703
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.628534509461326e-05,
+      "loss": 1.6625,
+      "step": 7704
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.626641683398127e-05,
+      "loss": 1.2588,
+      "step": 7705
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6247498605878697e-05,
+      "loss": 1.5283,
+      "step": 7706
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6228590412572264e-05,
+      "loss": 1.0426,
+      "step": 7707
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6209692256327413e-05,
+      "loss": 1.6367,
+      "step": 7708
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6190804139408465e-05,
+      "loss": 1.5448,
+      "step": 7709
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6171926064078467e-05,
+      "loss": 1.4659,
+      "step": 7710
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.615305803259933e-05,
+      "loss": 1.3389,
+      "step": 7711
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6134200047231674e-05,
+      "loss": 1.3535,
+      "step": 7712
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6115352110235026e-05,
+      "loss": 1.7796,
+      "step": 7713
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.609651422386761e-05,
+      "loss": 1.4189,
+      "step": 7714
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6077686390386503e-05,
+      "loss": 1.2643,
+      "step": 7715
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6058868612047517e-05,
+      "loss": 1.3119,
+      "step": 7716
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.604006089110537e-05,
+      "loss": 1.5691,
+      "step": 7717
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6021263229813443e-05,
+      "loss": 1.2853,
+      "step": 7718
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.600247563042403e-05,
+      "loss": 1.2237,
+      "step": 7719
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5983698095188138e-05,
+      "loss": 1.0793,
+      "step": 7720
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5964930626355568e-05,
+      "loss": 1.276,
+      "step": 7721
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5946173226174998e-05,
+      "loss": 1.4749,
+      "step": 7722
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.59274258968938e-05,
+      "loss": 1.2885,
+      "step": 7723
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5908688640758218e-05,
+      "loss": 1.3203,
+      "step": 7724
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5889961460013214e-05,
+      "loss": 1.2968,
+      "step": 7725
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5871244356902626e-05,
+      "loss": 1.3684,
+      "step": 7726
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5852537333669025e-05,
+      "loss": 1.1792,
+      "step": 7727
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.583384039255379e-05,
+      "loss": 1.4042,
+      "step": 7728
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5815153535797056e-05,
+      "loss": 1.5548,
+      "step": 7729
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5796476765637846e-05,
+      "loss": 1.404,
+      "step": 7730
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.577781008431386e-05,
+      "loss": 1.3972,
+      "step": 7731
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5759153494061705e-05,
+      "loss": 1.0695,
+      "step": 7732
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5740506997116688e-05,
+      "loss": 1.5459,
+      "step": 7733
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.572187059571293e-05,
+      "loss": 1.5258,
+      "step": 7734
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.570324429208333e-05,
+      "loss": 1.4504,
+      "step": 7735
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.568462808845964e-05,
+      "loss": 1.0018,
+      "step": 7736
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.566602198707232e-05,
+      "loss": 1.2081,
+      "step": 7737
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.56474259901507e-05,
+      "loss": 1.2846,
+      "step": 7738
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5628840099922815e-05,
+      "loss": 1.2922,
+      "step": 7739
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5610264318615597e-05,
+      "loss": 0.8967,
+      "step": 7740
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.55916986484546e-05,
+      "loss": 1.4402,
+      "step": 7741
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5573143091664343e-05,
+      "loss": 1.2435,
+      "step": 7742
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5554597650468015e-05,
+      "loss": 1.2729,
+      "step": 7743
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5536062327087686e-05,
+      "loss": 0.9299,
+      "step": 7744
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.551753712374412e-05,
+      "loss": 1.3576,
+      "step": 7745
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.549902204265693e-05,
+      "loss": 1.272,
+      "step": 7746
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.548051708604451e-05,
+      "loss": 1.2636,
+      "step": 7747
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.546202225612401e-05,
+      "loss": 1.4104,
+      "step": 7748
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5443537555111363e-05,
+      "loss": 1.2312,
+      "step": 7749
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5425062985221374e-05,
+      "loss": 1.2153,
+      "step": 7750
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5406598548667506e-05,
+      "loss": 1.3912,
+      "step": 7751
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5388144247662117e-05,
+      "loss": 1.1013,
+      "step": 7752
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5369700084416306e-05,
+      "loss": 1.1737,
+      "step": 7753
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5351266061139946e-05,
+      "loss": 1.1196,
+      "step": 7754
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5332842180041672e-05,
+      "loss": 1.2042,
+      "step": 7755
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5314428443328988e-05,
+      "loss": 1.4211,
+      "step": 7756
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5296024853208135e-05,
+      "loss": 1.2135,
+      "step": 7757
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5277631411884098e-05,
+      "loss": 1.2699,
+      "step": 7758
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5259248121560743e-05,
+      "loss": 1.4825,
+      "step": 7759
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5240874984440611e-05,
+      "loss": 1.3668,
+      "step": 7760
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5222512002725108e-05,
+      "loss": 1.4581,
+      "step": 7761
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5204159178614341e-05,
+      "loss": 1.6556,
+      "step": 7762
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5185816514307328e-05,
+      "loss": 1.211,
+      "step": 7763
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.516748401200172e-05,
+      "loss": 1.2801,
+      "step": 7764
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5149161673894086e-05,
+      "loss": 1.4058,
+      "step": 7765
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5130849502179678e-05,
+      "loss": 1.16,
+      "step": 7766
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5112547499052587e-05,
+      "loss": 1.3385,
+      "step": 7767
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5094255666705625e-05,
+      "loss": 1.2028,
+      "step": 7768
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5075974007330474e-05,
+      "loss": 1.1774,
+      "step": 7769
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5057702523117511e-05,
+      "loss": 1.3107,
+      "step": 7770
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5039441216255967e-05,
+      "loss": 1.0537,
+      "step": 7771
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.502119008893378e-05,
+      "loss": 1.1583,
+      "step": 7772
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5002949143337752e-05,
+      "loss": 1.4145,
+      "step": 7773
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4984718381653395e-05,
+      "loss": 1.2214,
+      "step": 7774
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4966497806065028e-05,
+      "loss": 1.2063,
+      "step": 7775
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4948287418755725e-05,
+      "loss": 1.2132,
+      "step": 7776
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4930087221907407e-05,
+      "loss": 1.2118,
+      "step": 7777
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4911897217700665e-05,
+      "loss": 1.1626,
+      "step": 7778
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4893717408315011e-05,
+      "loss": 0.8642,
+      "step": 7779
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4875547795928602e-05,
+      "loss": 1.4376,
+      "step": 7780
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.485738838271845e-05,
+      "loss": 1.2944,
+      "step": 7781
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4839239170860286e-05,
+      "loss": 1.6473,
+      "step": 7782
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4821100162528712e-05,
+      "loss": 1.2114,
+      "step": 7783
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4802971359897e-05,
+      "loss": 1.2493,
+      "step": 7784
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4784852765137303e-05,
+      "loss": 0.8356,
+      "step": 7785
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.476674438042044e-05,
+      "loss": 1.5481,
+      "step": 7786
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.474864620791614e-05,
+      "loss": 1.5011,
+      "step": 7787
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4730558249792748e-05,
+      "loss": 1.124,
+      "step": 7788
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.471248050821753e-05,
+      "loss": 1.2973,
+      "step": 7789
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.469441298535642e-05,
+      "loss": 1.2396,
+      "step": 7790
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4676355683374233e-05,
+      "loss": 1.0451,
+      "step": 7791
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4658308604434456e-05,
+      "loss": 1.2437,
+      "step": 7792
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.464027175069943e-05,
+      "loss": 1.174,
+      "step": 7793
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4622245124330235e-05,
+      "loss": 1.4272,
+      "step": 7794
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4604228727486691e-05,
+      "loss": 0.8559,
+      "step": 7795
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4586222562327479e-05,
+      "loss": 1.1991,
+      "step": 7796
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4568226631009962e-05,
+      "loss": 1.0744,
+      "step": 7797
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4550240935690374e-05,
+      "loss": 1.2686,
+      "step": 7798
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.453226547852361e-05,
+      "loss": 1.537,
+      "step": 7799
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.451430026166347e-05,
+      "loss": 1.3289,
+      "step": 7800
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.449634528726237e-05,
+      "loss": 1.2591,
+      "step": 7801
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.447840055747164e-05,
+      "loss": 1.5213,
+      "step": 7802
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4460466074441293e-05,
+      "loss": 1.1612,
+      "step": 7803
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4442541840320179e-05,
+      "loss": 1.4241,
+      "step": 7804
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4424627857255845e-05,
+      "loss": 1.2088,
+      "step": 7805
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.440672412739471e-05,
+      "loss": 1.1295,
+      "step": 7806
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.438883065288188e-05,
+      "loss": 0.9063,
+      "step": 7807
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4370947435861249e-05,
+      "loss": 1.5036,
+      "step": 7808
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4353074478475482e-05,
+      "loss": 1.2638,
+      "step": 7809
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4335211782866054e-05,
+      "loss": 1.1123,
+      "step": 7810
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4317359351173166e-05,
+      "loss": 1.3028,
+      "step": 7811
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4299517185535815e-05,
+      "loss": 1.4125,
+      "step": 7812
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4281685288091773e-05,
+      "loss": 1.4591,
+      "step": 7813
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4263863660977528e-05,
+      "loss": 1.5588,
+      "step": 7814
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4246052306328384e-05,
+      "loss": 1.0707,
+      "step": 7815
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4228251226278433e-05,
+      "loss": 1.2698,
+      "step": 7816
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.421046042296048e-05,
+      "loss": 1.211,
+      "step": 7817
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4192679898506157e-05,
+      "loss": 1.0592,
+      "step": 7818
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4174909655045798e-05,
+      "loss": 1.3498,
+      "step": 7819
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4157149694708594e-05,
+      "loss": 1.0568,
+      "step": 7820
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4139400019622418e-05,
+      "loss": 1.6396,
+      "step": 7821
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4121660631913946e-05,
+      "loss": 1.1672,
+      "step": 7822
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4103931533708603e-05,
+      "loss": 1.4353,
+      "step": 7823
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4086212727130655e-05,
+      "loss": 1.2053,
+      "step": 7824
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4068504214303014e-05,
+      "loss": 1.2042,
+      "step": 7825
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4050805997347471e-05,
+      "loss": 1.2457,
+      "step": 7826
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4033118078384523e-05,
+      "loss": 1.5347,
+      "step": 7827
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4015440459533425e-05,
+      "loss": 1.2304,
+      "step": 7828
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3997773142912252e-05,
+      "loss": 0.8077,
+      "step": 7829
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3980116130637777e-05,
+      "loss": 1.6467,
+      "step": 7830
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3962469424825608e-05,
+      "loss": 1.2941,
+      "step": 7831
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3944833027590043e-05,
+      "loss": 0.8583,
+      "step": 7832
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3927206941044235e-05,
+      "loss": 1.3737,
+      "step": 7833
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3909591167300029e-05,
+      "loss": 1.4044,
+      "step": 7834
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3891985708468048e-05,
+      "loss": 1.3198,
+      "step": 7835
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3874390566657669e-05,
+      "loss": 1.4759,
+      "step": 7836
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3856805743977098e-05,
+      "loss": 0.8341,
+      "step": 7837
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3839231242533201e-05,
+      "loss": 1.2918,
+      "step": 7838
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3821667064431731e-05,
+      "loss": 1.2717,
+      "step": 7839
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3804113211777115e-05,
+      "loss": 1.3019,
+      "step": 7840
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3786569686672546e-05,
+      "loss": 1.2789,
+      "step": 7841
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3769036491219988e-05,
+      "loss": 1.1532,
+      "step": 7842
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.375151362752023e-05,
+      "loss": 1.1034,
+      "step": 7843
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3734001097672721e-05,
+      "loss": 1.6829,
+      "step": 7844
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3716498903775765e-05,
+      "loss": 1.3673,
+      "step": 7845
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3699007047926338e-05,
+      "loss": 1.1111,
+      "step": 7846
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3681525532220296e-05,
+      "loss": 1.3425,
+      "step": 7847
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3664054358752098e-05,
+      "loss": 1.4047,
+      "step": 7848
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3646593529615126e-05,
+      "loss": 1.3232,
+      "step": 7849
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3629143046901382e-05,
+      "loss": 1.0032,
+      "step": 7850
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.361170291270174e-05,
+      "loss": 1.5208,
+      "step": 7851
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3594273129105762e-05,
+      "loss": 1.323,
+      "step": 7852
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.357685369820183e-05,
+      "loss": 0.7307,
+      "step": 7853
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3559444622077022e-05,
+      "loss": 1.364,
+      "step": 7854
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3542045902817212e-05,
+      "loss": 1.6953,
+      "step": 7855
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3524657542507002e-05,
+      "loss": 1.2878,
+      "step": 7856
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3507279543229823e-05,
+      "loss": 1.2276,
+      "step": 7857
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3489911907067776e-05,
+      "loss": 1.191,
+      "step": 7858
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3472554636101797e-05,
+      "loss": 1.2541,
+      "step": 7859
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3455207732411523e-05,
+      "loss": 1.3073,
+      "step": 7860
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.343787119807538e-05,
+      "loss": 1.297,
+      "step": 7861
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3420545035170517e-05,
+      "loss": 1.3991,
+      "step": 7862
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3403229245772908e-05,
+      "loss": 1.2833,
+      "step": 7863
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3385923831957192e-05,
+      "loss": 1.3498,
+      "step": 7864
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3368628795796855e-05,
+      "loss": 1.2201,
+      "step": 7865
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3351344139364119e-05,
+      "loss": 1.2526,
+      "step": 7866
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3334069864729903e-05,
+      "loss": 1.1142,
+      "step": 7867
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3316805973963941e-05,
+      "loss": 0.9604,
+      "step": 7868
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3299552469134669e-05,
+      "loss": 1.1033,
+      "step": 7869
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3282309352309363e-05,
+      "loss": 1.1979,
+      "step": 7870
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3265076625553963e-05,
+      "loss": 1.2582,
+      "step": 7871
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3247854290933249e-05,
+      "loss": 1.4127,
+      "step": 7872
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3230642350510692e-05,
+      "loss": 1.2993,
+      "step": 7873
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.321344080634853e-05,
+      "loss": 1.4676,
+      "step": 7874
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3196249660507764e-05,
+      "loss": 1.1452,
+      "step": 7875
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3179068915048165e-05,
+      "loss": 1.4047,
+      "step": 7876
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3161898572028208e-05,
+      "loss": 1.2536,
+      "step": 7877
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3144738633505217e-05,
+      "loss": 1.384,
+      "step": 7878
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3127589101535143e-05,
+      "loss": 0.8314,
+      "step": 7879
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.31104499781728e-05,
+      "loss": 1.2461,
+      "step": 7880
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3093321265471692e-05,
+      "loss": 1.2112,
+      "step": 7881
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3076202965484107e-05,
+      "loss": 1.4513,
+      "step": 7882
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.305909508026103e-05,
+      "loss": 1.3023,
+      "step": 7883
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3041997611852296e-05,
+      "loss": 1.2084,
+      "step": 7884
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3024910562306392e-05,
+      "loss": 1.2072,
+      "step": 7885
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3007833933670643e-05,
+      "loss": 1.431,
+      "step": 7886
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.299076772799105e-05,
+      "loss": 1.2724,
+      "step": 7887
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2973711947312416e-05,
+      "loss": 1.3741,
+      "step": 7888
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2956666593678247e-05,
+      "loss": 1.1603,
+      "step": 7889
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.293963166913088e-05,
+      "loss": 1.5638,
+      "step": 7890
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.29226071757113e-05,
+      "loss": 1.1748,
+      "step": 7891
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2905593115459335e-05,
+      "loss": 1.3971,
+      "step": 7892
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2888589490413495e-05,
+      "loss": 1.6284,
+      "step": 7893
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.28715963026111e-05,
+      "loss": 1.1926,
+      "step": 7894
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2854613554088169e-05,
+      "loss": 1.3382,
+      "step": 7895
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2837641246879483e-05,
+      "loss": 1.4533,
+      "step": 7896
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2820679383018563e-05,
+      "loss": 1.2329,
+      "step": 7897
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2803727964537726e-05,
+      "loss": 1.5089,
+      "step": 7898
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.278678699346797e-05,
+      "loss": 1.2614,
+      "step": 7899
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2769856471839103e-05,
+      "loss": 1.2455,
+      "step": 7900
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.275293640167966e-05,
+      "loss": 1.4387,
+      "step": 7901
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2736026785016875e-05,
+      "loss": 1.2937,
+      "step": 7902
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2719127623876814e-05,
+      "loss": 1.2849,
+      "step": 7903
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2702238920284215e-05,
+      "loss": 1.5345,
+      "step": 7904
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2685360676262636e-05,
+      "loss": 1.6563,
+      "step": 7905
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2668492893834295e-05,
+      "loss": 1.3582,
+      "step": 7906
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.265163557502026e-05,
+      "loss": 1.4228,
+      "step": 7907
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.263478872184023e-05,
+      "loss": 1.2557,
+      "step": 7908
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2617952336312743e-05,
+      "loss": 1.38,
+      "step": 7909
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2601126420455024e-05,
+      "loss": 1.2755,
+      "step": 7910
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2584310976283109e-05,
+      "loss": 1.2873,
+      "step": 7911
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.256750600581169e-05,
+      "loss": 1.2596,
+      "step": 7912
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2550711511054291e-05,
+      "loss": 1.2075,
+      "step": 7913
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2533927494023135e-05,
+      "loss": 1.3689,
+      "step": 7914
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2517153956729188e-05,
+      "loss": 1.1229,
+      "step": 7915
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2500390901182147e-05,
+      "loss": 1.2749,
+      "step": 7916
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2483638329390523e-05,
+      "loss": 1.6714,
+      "step": 7917
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2466896243361482e-05,
+      "loss": 1.3711,
+      "step": 7918
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2450164645101014e-05,
+      "loss": 1.2293,
+      "step": 7919
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2433443536613786e-05,
+      "loss": 1.0709,
+      "step": 7920
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2416732919903251e-05,
+      "loss": 1.3953,
+      "step": 7921
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2400032796971551e-05,
+      "loss": 1.3434,
+      "step": 7922
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2383343169819662e-05,
+      "loss": 1.242,
+      "step": 7923
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2366664040447217e-05,
+      "loss": 1.2014,
+      "step": 7924
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2349995410852643e-05,
+      "loss": 1.2519,
+      "step": 7925
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2333337283033065e-05,
+      "loss": 1.5667,
+      "step": 7926
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2316689658984404e-05,
+      "loss": 1.6376,
+      "step": 7927
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.230005254070129e-05,
+      "loss": 1.4038,
+      "step": 7928
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2283425930177073e-05,
+      "loss": 1.0799,
+      "step": 7929
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.226680982940387e-05,
+      "loss": 1.1577,
+      "step": 7930
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2250204240372576e-05,
+      "loss": 1.2507,
+      "step": 7931
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2233609165072724e-05,
+      "loss": 1.2521,
+      "step": 7932
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2217024605492721e-05,
+      "loss": 1.4303,
+      "step": 7933
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2200450563619614e-05,
+      "loss": 1.6323,
+      "step": 7934
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2183887041439213e-05,
+      "loss": 1.3235,
+      "step": 7935
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2167334040936051e-05,
+      "loss": 0.9615,
+      "step": 7936
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2150791564093456e-05,
+      "loss": 1.4081,
+      "step": 7937
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2134259612893484e-05,
+      "loss": 1.2663,
+      "step": 7938
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2117738189316853e-05,
+      "loss": 1.2248,
+      "step": 7939
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2101227295343132e-05,
+      "loss": 1.3836,
+      "step": 7940
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.208472693295054e-05,
+      "loss": 1.1203,
+      "step": 7941
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2068237104116076e-05,
+      "loss": 1.3308,
+      "step": 7942
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2051757810815444e-05,
+      "loss": 1.4165,
+      "step": 7943
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2035289055023136e-05,
+      "loss": 1.1934,
+      "step": 7944
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2018830838712336e-05,
+      "loss": 1.1812,
+      "step": 7945
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2002383163855012e-05,
+      "loss": 1.2851,
+      "step": 7946
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1985946032421813e-05,
+      "loss": 1.558,
+      "step": 7947
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.196951944638216e-05,
+      "loss": 1.3411,
+      "step": 7948
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1953103407704192e-05,
+      "loss": 1.2106,
+      "step": 7949
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1936697918354811e-05,
+      "loss": 1.7083,
+      "step": 7950
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1920302980299613e-05,
+      "loss": 1.5261,
+      "step": 7951
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1903918595503006e-05,
+      "loss": 1.002,
+      "step": 7952
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1887544765928028e-05,
+      "loss": 1.2629,
+      "step": 7953
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1871181493536542e-05,
+      "loss": 1.1708,
+      "step": 7954
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1854828780289106e-05,
+      "loss": 1.2161,
+      "step": 7955
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1838486628145018e-05,
+      "loss": 1.3801,
+      "step": 7956
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1822155039062288e-05,
+      "loss": 1.4727,
+      "step": 7957
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1805834014997719e-05,
+      "loss": 1.7062,
+      "step": 7958
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1789523557906768e-05,
+      "loss": 1.1507,
+      "step": 7959
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1773223669743728e-05,
+      "loss": 1.2288,
+      "step": 7960
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1756934352461535e-05,
+      "loss": 1.6332,
+      "step": 7961
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1740655608011886e-05,
+      "loss": 1.4756,
+      "step": 7962
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1724387438345219e-05,
+      "loss": 1.2128,
+      "step": 7963
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.170812984541072e-05,
+      "loss": 0.8703,
+      "step": 7964
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1691882831156265e-05,
+      "loss": 1.1991,
+      "step": 7965
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1675646397528517e-05,
+      "loss": 1.415,
+      "step": 7966
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1659420546472832e-05,
+      "loss": 1.2475,
+      "step": 7967
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.164320527993331e-05,
+      "loss": 1.1798,
+      "step": 7968
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1627000599852755e-05,
+      "loss": 1.3762,
+      "step": 7969
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1610806508172767e-05,
+      "loss": 1.1819,
+      "step": 7970
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1594623006833605e-05,
+      "loss": 1.284,
+      "step": 7971
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1578450097774307e-05,
+      "loss": 1.0491,
+      "step": 7972
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1562287782932645e-05,
+      "loss": 1.1634,
+      "step": 7973
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1546136064245106e-05,
+      "loss": 1.3461,
+      "step": 7974
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1529994943646882e-05,
+      "loss": 0.7974,
+      "step": 7975
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1513864423071918e-05,
+      "loss": 1.198,
+      "step": 7976
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1497744504452912e-05,
+      "loss": 1.5172,
+      "step": 7977
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1481635189721251e-05,
+      "loss": 1.1939,
+      "step": 7978
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1465536480807093e-05,
+      "loss": 1.0811,
+      "step": 7979
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1449448379639283e-05,
+      "loss": 1.3729,
+      "step": 7980
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1433370888145434e-05,
+      "loss": 1.2867,
+      "step": 7981
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1417304008251828e-05,
+      "loss": 1.5138,
+      "step": 7982
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.140124774188357e-05,
+      "loss": 1.2132,
+      "step": 7983
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1385202090964386e-05,
+      "loss": 1.2222,
+      "step": 7984
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.136916705741684e-05,
+      "loss": 1.4888,
+      "step": 7985
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1353142643162118e-05,
+      "loss": 1.4434,
+      "step": 7986
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1337128850120227e-05,
+      "loss": 1.1074,
+      "step": 7987
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1321125680209833e-05,
+      "loss": 1.4993,
+      "step": 7988
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1305133135348355e-05,
+      "loss": 1.1084,
+      "step": 7989
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.128915121745192e-05,
+      "loss": 1.3131,
+      "step": 7990
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.127317992843544e-05,
+      "loss": 1.1951,
+      "step": 7991
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1257219270212472e-05,
+      "loss": 1.787,
+      "step": 7992
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1241269244695374e-05,
+      "loss": 0.8861,
+      "step": 7993
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1225329853795196e-05,
+      "loss": 1.1391,
+      "step": 7994
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.120940109942169e-05,
+      "loss": 0.8557,
+      "step": 7995
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1193482983483361e-05,
+      "loss": 1.2004,
+      "step": 7996
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.117757550788745e-05,
+      "loss": 1.3792,
+      "step": 7997
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1161678674539899e-05,
+      "loss": 0.9914,
+      "step": 7998
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1145792485345397e-05,
+      "loss": 1.2796,
+      "step": 7999
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.112991694220733e-05,
+      "loss": 1.4808,
+      "step": 8000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1114052047027845e-05,
+      "loss": 1.1696,
+      "step": 8001
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1098197801707787e-05,
+      "loss": 1.3188,
+      "step": 8002
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1082354208146717e-05,
+      "loss": 0.9936,
+      "step": 8003
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1066521268242924e-05,
+      "loss": 1.4321,
+      "step": 8004
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.105069898389347e-05,
+      "loss": 1.4882,
+      "step": 8005
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1034887356994051e-05,
+      "loss": 1.3973,
+      "step": 8006
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1019086389439182e-05,
+      "loss": 1.4612,
+      "step": 8007
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1003296083122027e-05,
+      "loss": 1.3085,
+      "step": 8008
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0987516439934498e-05,
+      "loss": 1.3565,
+      "step": 8009
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0971747461767257e-05,
+      "loss": 0.998,
+      "step": 8010
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0955989150509616e-05,
+      "loss": 1.5428,
+      "step": 8011
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0940241508049709e-05,
+      "loss": 1.1967,
+      "step": 8012
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0924504536274283e-05,
+      "loss": 1.334,
+      "step": 8013
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0908778237068917e-05,
+      "loss": 1.4872,
+      "step": 8014
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0893062612317827e-05,
+      "loss": 1.4098,
+      "step": 8015
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0877357663903986e-05,
+      "loss": 1.2138,
+      "step": 8016
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0861663393709054e-05,
+      "loss": 1.1881,
+      "step": 8017
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0845979803613481e-05,
+      "loss": 1.4899,
+      "step": 8018
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0830306895496346e-05,
+      "loss": 1.3048,
+      "step": 8019
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0814644671235552e-05,
+      "loss": 1.2513,
+      "step": 8020
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0798993132707635e-05,
+      "loss": 1.349,
+      "step": 8021
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0783352281787885e-05,
+      "loss": 1.432,
+      "step": 8022
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0767722120350287e-05,
+      "loss": 1.263,
+      "step": 8023
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0752102650267614e-05,
+      "loss": 1.1377,
+      "step": 8024
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0736493873411258e-05,
+      "loss": 1.7343,
+      "step": 8025
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0720895791651442e-05,
+      "loss": 1.082,
+      "step": 8026
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0705308406857007e-05,
+      "loss": 1.3112,
+      "step": 8027
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0689731720895569e-05,
+      "loss": 1.4702,
+      "step": 8028
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0674165735633423e-05,
+      "loss": 1.2326,
+      "step": 8029
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0658610452935646e-05,
+      "loss": 1.3637,
+      "step": 8030
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0643065874665948e-05,
+      "loss": 1.6417,
+      "step": 8031
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0627532002686835e-05,
+      "loss": 1.7383,
+      "step": 8032
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0612008838859477e-05,
+      "loss": 1.3845,
+      "step": 8033
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0596496385043809e-05,
+      "loss": 1.3798,
+      "step": 8034
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0580994643098418e-05,
+      "loss": 1.1429,
+      "step": 8035
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0565503614880667e-05,
+      "loss": 1.0002,
+      "step": 8036
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0550023302246592e-05,
+      "loss": 1.4018,
+      "step": 8037
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0534553707050987e-05,
+      "loss": 1.3242,
+      "step": 8038
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0519094831147303e-05,
+      "loss": 1.1984,
+      "step": 8039
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.050364667638779e-05,
+      "loss": 0.8257,
+      "step": 8040
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0488209244623348e-05,
+      "loss": 1.3517,
+      "step": 8041
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0472782537703618e-05,
+      "loss": 0.8306,
+      "step": 8042
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.045736655747691e-05,
+      "loss": 1.219,
+      "step": 8043
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0441961305790315e-05,
+      "loss": 1.2254,
+      "step": 8044
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0426566784489644e-05,
+      "loss": 1.3085,
+      "step": 8045
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0411182995419334e-05,
+      "loss": 1.3508,
+      "step": 8046
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.039580994042263e-05,
+      "loss": 1.0819,
+      "step": 8047
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0380447621341449e-05,
+      "loss": 1.4095,
+      "step": 8048
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.036509604001642e-05,
+      "loss": 1.3389,
+      "step": 8049
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.034975519828686e-05,
+      "loss": 1.0314,
+      "step": 8050
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0334425097990886e-05,
+      "loss": 1.2633,
+      "step": 8051
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0319105740965218e-05,
+      "loss": 1.3247,
+      "step": 8052
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.030379712904539e-05,
+      "loss": 1.366,
+      "step": 8053
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0288499264065576e-05,
+      "loss": 1.5724,
+      "step": 8054
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0273212147858701e-05,
+      "loss": 1.4985,
+      "step": 8055
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0257935782256356e-05,
+      "loss": 1.4538,
+      "step": 8056
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.024267016908893e-05,
+      "loss": 1.1701,
+      "step": 8057
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0227415310185406e-05,
+      "loss": 1.2071,
+      "step": 8058
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.021217120737361e-05,
+      "loss": 1.0102,
+      "step": 8059
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.019693786247996e-05,
+      "loss": 0.9375,
+      "step": 8060
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0181715277329684e-05,
+      "loss": 1.2927,
+      "step": 8061
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0166503453746646e-05,
+      "loss": 1.1455,
+      "step": 8062
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0151302393553441e-05,
+      "loss": 1.4823,
+      "step": 8063
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0136112098571381e-05,
+      "loss": 1.3352,
+      "step": 8064
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0120932570620534e-05,
+      "loss": 1.2903,
+      "step": 8065
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0105763811519564e-05,
+      "loss": 1.1253,
+      "step": 8066
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0090605823085974e-05,
+      "loss": 1.4837,
+      "step": 8067
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0075458607135902e-05,
+      "loss": 1.0726,
+      "step": 8068
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0060322165484205e-05,
+      "loss": 1.6112,
+      "step": 8069
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.004519649994442e-05,
+      "loss": 1.1767,
+      "step": 8070
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0030081612328878e-05,
+      "loss": 1.0138,
+      "step": 8071
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.001497750444853e-05,
+      "loss": 1.3992,
+      "step": 8072
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.999884178113106e-06,
+      "loss": 1.1378,
+      "step": 8073
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.98480163513098e-06,
+      "loss": 1.3173,
+      "step": 8074
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.96972987730932e-06,
+      "loss": 1.3399,
+      "step": 8075
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.954668906453868e-06,
+      "loss": 1.2789,
+      "step": 8076
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.939618724369215e-06,
+      "loss": 1.174,
+      "step": 8077
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.924579332858552e-06,
+      "loss": 1.3319,
+      "step": 8078
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.90955073372385e-06,
+      "loss": 1.2956,
+      "step": 8079
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.894532928765742e-06,
+      "loss": 1.0143,
+      "step": 8080
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.879525919783617e-06,
+      "loss": 1.179,
+      "step": 8081
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.86452970857552e-06,
+      "loss": 1.3672,
+      "step": 8082
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.849544296938184e-06,
+      "loss": 1.014,
+      "step": 8083
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.83456968666715e-06,
+      "loss": 1.2015,
+      "step": 8084
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.819605879556538e-06,
+      "loss": 1.2107,
+      "step": 8085
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.804652877399278e-06,
+      "loss": 1.5582,
+      "step": 8086
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.789710681986952e-06,
+      "loss": 1.3076,
+      "step": 8087
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.774779295109847e-06,
+      "loss": 1.6332,
+      "step": 8088
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.759858718556958e-06,
+      "loss": 1.0688,
+      "step": 8089
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.744948954116018e-06,
+      "loss": 1.0208,
+      "step": 8090
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.73005000357341e-06,
+      "loss": 1.2898,
+      "step": 8091
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.715161868714273e-06,
+      "loss": 1.4546,
+      "step": 8092
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.700284551322414e-06,
+      "loss": 1.6027,
+      "step": 8093
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.685418053180372e-06,
+      "loss": 1.3901,
+      "step": 8094
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.67056237606937e-06,
+      "loss": 1.0154,
+      "step": 8095
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.655717521769336e-06,
+      "loss": 0.7763,
+      "step": 8096
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.640883492058882e-06,
+      "loss": 1.493,
+      "step": 8097
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.626060288715388e-06,
+      "loss": 1.3578,
+      "step": 8098
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.611247913514843e-06,
+      "loss": 1.3265,
+      "step": 8099
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.596446368232049e-06,
+      "loss": 1.4667,
+      "step": 8100
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.58165565464042e-06,
+      "loss": 1.1229,
+      "step": 8101
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.566875774512097e-06,
+      "loss": 1.2463,
+      "step": 8102
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.552106729617927e-06,
+      "loss": 1.3424,
+      "step": 8103
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.537348521727497e-06,
+      "loss": 1.4056,
+      "step": 8104
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.522601152609012e-06,
+      "loss": 1.0676,
+      "step": 8105
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.507864624029461e-06,
+      "loss": 0.951,
+      "step": 8106
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.493138937754475e-06,
+      "loss": 1.1826,
+      "step": 8107
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.478424095548443e-06,
+      "loss": 1.0383,
+      "step": 8108
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.4637200991744e-06,
+      "loss": 1.575,
+      "step": 8109
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.449026950394113e-06,
+      "loss": 1.3185,
+      "step": 8110
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.434344650968007e-06,
+      "loss": 1.0343,
+      "step": 8111
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.419673202655287e-06,
+      "loss": 1.5629,
+      "step": 8112
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.405012607213781e-06,
+      "loss": 1.3213,
+      "step": 8113
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.39036286640006e-06,
+      "loss": 1.005,
+      "step": 8114
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.375723981969387e-06,
+      "loss": 1.3443,
+      "step": 8115
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.36109595567568e-06,
+      "loss": 1.2593,
+      "step": 8116
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.34647878927165e-06,
+      "loss": 1.0833,
+      "step": 8117
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.331872484508597e-06,
+      "loss": 1.5603,
+      "step": 8118
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.317277043136618e-06,
+      "loss": 1.6612,
+      "step": 8119
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.302692466904438e-06,
+      "loss": 1.2153,
+      "step": 8120
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.288118757559528e-06,
+      "loss": 1.298,
+      "step": 8121
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.273555916848009e-06,
+      "loss": 1.2708,
+      "step": 8122
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.259003946514755e-06,
+      "loss": 1.343,
+      "step": 8123
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.24446284830326e-06,
+      "loss": 1.0351,
+      "step": 8124
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.229932623955828e-06,
+      "loss": 1.1465,
+      "step": 8125
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.215413275213336e-06,
+      "loss": 1.6165,
+      "step": 8126
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.200904803815469e-06,
+      "loss": 1.1935,
+      "step": 8127
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.186407211500537e-06,
+      "loss": 1.5355,
+      "step": 8128
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.17192050000556e-06,
+      "loss": 1.1276,
+      "step": 8129
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.15744467106625e-06,
+      "loss": 1.1656,
+      "step": 8130
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.142979726417066e-06,
+      "loss": 1.0859,
+      "step": 8131
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.128525667791076e-06,
+      "loss": 1.2601,
+      "step": 8132
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.11408249692014e-06,
+      "loss": 1.6575,
+      "step": 8133
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.099650215534727e-06,
+      "loss": 1.2113,
+      "step": 8134
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.085228825364078e-06,
+      "loss": 1.6412,
+      "step": 8135
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.070818328136045e-06,
+      "loss": 1.4573,
+      "step": 8136
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.056418725577253e-06,
+      "loss": 1.1067,
+      "step": 8137
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.042030019412951e-06,
+      "loss": 1.3127,
+      "step": 8138
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.027652211367177e-06,
+      "loss": 1.2811,
+      "step": 8139
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.013285303162544e-06,
+      "loss": 1.2706,
+      "step": 8140
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.998929296520475e-06,
+      "loss": 1.3973,
+      "step": 8141
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.984584193161017e-06,
+      "loss": 1.0981,
+      "step": 8142
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.970249994802915e-06,
+      "loss": 1.444,
+      "step": 8143
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.955926703163608e-06,
+      "loss": 1.4857,
+      "step": 8144
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.941614319959268e-06,
+      "loss": 1.0349,
+      "step": 8145
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.927312846904711e-06,
+      "loss": 1.3769,
+      "step": 8146
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.913022285713491e-06,
+      "loss": 1.433,
+      "step": 8147
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.898742638097812e-06,
+      "loss": 1.2582,
+      "step": 8148
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.884473905768598e-06,
+      "loss": 1.0823,
+      "step": 8149
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.870216090435423e-06,
+      "loss": 1.4539,
+      "step": 8150
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.855969193806623e-06,
+      "loss": 1.1311,
+      "step": 8151
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.841733217589165e-06,
+      "loss": 0.997,
+      "step": 8152
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.82750816348874e-06,
+      "loss": 1.3676,
+      "step": 8153
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.81329403320975e-06,
+      "loss": 1.4213,
+      "step": 8154
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.799090828455214e-06,
+      "loss": 0.9594,
+      "step": 8155
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.784898550926923e-06,
+      "loss": 1.2443,
+      "step": 8156
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.770717202325273e-06,
+      "loss": 1.5506,
+      "step": 8157
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.756546784349462e-06,
+      "loss": 1.2151,
+      "step": 8158
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.742387298697264e-06,
+      "loss": 1.2003,
+      "step": 8159
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.728238747065242e-06,
+      "loss": 1.6255,
+      "step": 8160
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.714101131148577e-06,
+      "loss": 0.5933,
+      "step": 8161
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.699974452641168e-06,
+      "loss": 1.0845,
+      "step": 8162
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.685858713235595e-06,
+      "loss": 1.1946,
+      "step": 8163
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.671753914623148e-06,
+      "loss": 1.2163,
+      "step": 8164
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.657660058493777e-06,
+      "loss": 1.2014,
+      "step": 8165
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.643577146536163e-06,
+      "loss": 1.1952,
+      "step": 8166
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.629505180437602e-06,
+      "loss": 1.2723,
+      "step": 8167
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.615444161884179e-06,
+      "loss": 1.0798,
+      "step": 8168
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.601394092560588e-06,
+      "loss": 1.0166,
+      "step": 8169
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.58735497415023e-06,
+      "loss": 1.4208,
+      "step": 8170
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.57332680833519e-06,
+      "loss": 1.0255,
+      "step": 8171
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.559309596796295e-06,
+      "loss": 1.4148,
+      "step": 8172
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.545303341212951e-06,
+      "loss": 1.1827,
+      "step": 8173
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.531308043263386e-06,
+      "loss": 1.2461,
+      "step": 8174
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.517323704624403e-06,
+      "loss": 1.1986,
+      "step": 8175
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.503350326971549e-06,
+      "loss": 1.1037,
+      "step": 8176
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.489387911979007e-06,
+      "loss": 1.6072,
+      "step": 8177
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.47543646131974e-06,
+      "loss": 1.5471,
+      "step": 8178
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.46149597666529e-06,
+      "loss": 1.0342,
+      "step": 8179
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.447566459685963e-06,
+      "loss": 1.1486,
+      "step": 8180
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.433647912050712e-06,
+      "loss": 1.127,
+      "step": 8181
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.419740335427206e-06,
+      "loss": 1.3554,
+      "step": 8182
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.405843731481744e-06,
+      "loss": 1.2752,
+      "step": 8183
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.391958101879372e-06,
+      "loss": 0.9638,
+      "step": 8184
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.378083448283768e-06,
+      "loss": 1.4745,
+      "step": 8185
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.364219772357373e-06,
+      "loss": 1.1486,
+      "step": 8186
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.3503670757612e-06,
+      "loss": 1.0749,
+      "step": 8187
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.336525360155056e-06,
+      "loss": 1.1555,
+      "step": 8188
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.32269462719737e-06,
+      "loss": 1.3828,
+      "step": 8189
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.308874878545248e-06,
+      "loss": 1.2657,
+      "step": 8190
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.295066115854532e-06,
+      "loss": 1.2696,
+      "step": 8191
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.28126834077969e-06,
+      "loss": 1.2721,
+      "step": 8192
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.267481554973944e-06,
+      "loss": 1.1559,
+      "step": 8193
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.253705760089104e-06,
+      "loss": 1.1388,
+      "step": 8194
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.239940957775771e-06,
+      "loss": 1.3849,
+      "step": 8195
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.226187149683107e-06,
+      "loss": 1.0035,
+      "step": 8196
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.21244433745908e-06,
+      "loss": 1.2016,
+      "step": 8197
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.19871252275024e-06,
+      "loss": 1.2142,
+      "step": 8198
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.184991707201895e-06,
+      "loss": 1.1633,
+      "step": 8199
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.171281892457983e-06,
+      "loss": 1.2759,
+      "step": 8200
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.157583080161168e-06,
+      "loss": 1.3702,
+      "step": 8201
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.14389527195275e-06,
+      "loss": 1.1961,
+      "step": 8202
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.130218469472739e-06,
+      "loss": 1.1973,
+      "step": 8203
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.116552674359812e-06,
+      "loss": 1.4157,
+      "step": 8204
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.102897888251348e-06,
+      "loss": 1.2992,
+      "step": 8205
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.089254112783374e-06,
+      "loss": 1.1194,
+      "step": 8206
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.075621349590645e-06,
+      "loss": 1.4893,
+      "step": 8207
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.061999600306558e-06,
+      "loss": 1.3683,
+      "step": 8208
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.048388866563194e-06,
+      "loss": 1.1601,
+      "step": 8209
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.034789149991306e-06,
+      "loss": 1.1572,
+      "step": 8210
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.02120045222038e-06,
+      "loss": 1.0729,
+      "step": 8211
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.00762277487851e-06,
+      "loss": 1.2836,
+      "step": 8212
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.994056119592541e-06,
+      "loss": 1.2932,
+      "step": 8213
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.980500487987908e-06,
+      "loss": 0.8746,
+      "step": 8214
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.966955881688842e-06,
+      "loss": 1.2774,
+      "step": 8215
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.953422302318137e-06,
+      "loss": 0.843,
+      "step": 8216
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.939899751497349e-06,
+      "loss": 1.4394,
+      "step": 8217
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.926388230846637e-06,
+      "loss": 1.0432,
+      "step": 8218
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.912887741984942e-06,
+      "loss": 1.2133,
+      "step": 8219
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.89939828652977e-06,
+      "loss": 1.191,
+      "step": 8220
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.885919866097391e-06,
+      "loss": 1.4516,
+      "step": 8221
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.872452482302717e-06,
+      "loss": 1.6434,
+      "step": 8222
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.858996136759344e-06,
+      "loss": 1.1672,
+      "step": 8223
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.845550831079507e-06,
+      "loss": 1.3526,
+      "step": 8224
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.832116566874182e-06,
+      "loss": 1.4416,
+      "step": 8225
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.818693345753004e-06,
+      "loss": 1.6056,
+      "step": 8226
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.805281169324252e-06,
+      "loss": 0.7772,
+      "step": 8227
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.791880039194933e-06,
+      "loss": 1.1619,
+      "step": 8228
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.77848995697068e-06,
+      "loss": 1.251,
+      "step": 8229
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.765110924255825e-06,
+      "loss": 1.3912,
+      "step": 8230
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.751742942653372e-06,
+      "loss": 1.371,
+      "step": 8231
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.73838601376502e-06,
+      "loss": 1.0592,
+      "step": 8232
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.725040139191097e-06,
+      "loss": 1.4967,
+      "step": 8233
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.711705320530682e-06,
+      "loss": 0.996,
+      "step": 8234
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.698381559381462e-06,
+      "loss": 1.3664,
+      "step": 8235
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.685068857339816e-06,
+      "loss": 1.2823,
+      "step": 8236
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.671767216000803e-06,
+      "loss": 1.3159,
+      "step": 8237
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.658476636958167e-06,
+      "loss": 1.4261,
+      "step": 8238
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.645197121804304e-06,
+      "loss": 1.3073,
+      "step": 8239
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.631928672130329e-06,
+      "loss": 1.395,
+      "step": 8240
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.618671289525958e-06,
+      "loss": 0.9783,
+      "step": 8241
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.605424975579667e-06,
+      "loss": 1.2449,
+      "step": 8242
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.592189731878518e-06,
+      "loss": 1.3775,
+      "step": 8243
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.578965560008311e-06,
+      "loss": 1.2151,
+      "step": 8244
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.5657524615534885e-06,
+      "loss": 1.1735,
+      "step": 8245
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.552550438097206e-06,
+      "loss": 1.4091,
+      "step": 8246
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.53935949122122e-06,
+      "loss": 1.3936,
+      "step": 8247
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.526179622506046e-06,
+      "loss": 1.2798,
+      "step": 8248
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.513010833530798e-06,
+      "loss": 1.2667,
+      "step": 8249
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.499853125873302e-06,
+      "loss": 1.1769,
+      "step": 8250
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.48670650111003e-06,
+      "loss": 1.4141,
+      "step": 8251
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.4735709608161785e-06,
+      "loss": 1.6151,
+      "step": 8252
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.460446506565555e-06,
+      "loss": 1.4774,
+      "step": 8253
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.447333139930668e-06,
+      "loss": 1.3723,
+      "step": 8254
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.434230862482716e-06,
+      "loss": 1.0986,
+      "step": 8255
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.421139675791511e-06,
+      "loss": 1.1918,
+      "step": 8256
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.408059581425586e-06,
+      "loss": 1.0677,
+      "step": 8257
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.3949905809521434e-06,
+      "loss": 1.6657,
+      "step": 8258
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.38193267593702e-06,
+      "loss": 1.4115,
+      "step": 8259
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.368885867944753e-06,
+      "loss": 1.3546,
+      "step": 8260
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.35585015853858e-06,
+      "loss": 1.0788,
+      "step": 8261
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.342825549280341e-06,
+      "loss": 1.0129,
+      "step": 8262
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.329812041730588e-06,
+      "loss": 1.3912,
+      "step": 8263
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.3168096374485074e-06,
+      "loss": 1.4335,
+      "step": 8264
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.303818337992019e-06,
+      "loss": 1.5108,
+      "step": 8265
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.290838144917644e-06,
+      "loss": 1.2541,
+      "step": 8266
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.277869059780618e-06,
+      "loss": 0.9376,
+      "step": 8267
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.26491108413484e-06,
+      "loss": 1.3519,
+      "step": 8268
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.251964219532847e-06,
+      "loss": 1.2866,
+      "step": 8269
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.239028467525866e-06,
+      "loss": 1.4259,
+      "step": 8270
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.226103829663811e-06,
+      "loss": 1.1061,
+      "step": 8271
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.2131903074952215e-06,
+      "loss": 1.4138,
+      "step": 8272
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.2002879025673595e-06,
+      "loss": 1.0968,
+      "step": 8273
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.187396616426101e-06,
+      "loss": 1.4351,
+      "step": 8274
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.174516450616031e-06,
+      "loss": 1.3663,
+      "step": 8275
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.161647406680372e-06,
+      "loss": 1.1932,
+      "step": 8276
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.148789486161045e-06,
+      "loss": 1.4259,
+      "step": 8277
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.135942690598585e-06,
+      "loss": 1.3864,
+      "step": 8278
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.123107021532261e-06,
+      "loss": 1.2217,
+      "step": 8279
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.110282480499952e-06,
+      "loss": 0.8293,
+      "step": 8280
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.097469069038265e-06,
+      "loss": 1.6307,
+      "step": 8281
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.084666788682426e-06,
+      "loss": 1.0312,
+      "step": 8282
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.071875640966319e-06,
+      "loss": 1.2785,
+      "step": 8283
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.059095627422507e-06,
+      "loss": 1.2414,
+      "step": 8284
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.046326749582266e-06,
+      "loss": 1.0919,
+      "step": 8285
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.033569008975449e-06,
+      "loss": 1.3242,
+      "step": 8286
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.020822407130667e-06,
+      "loss": 1.3432,
+      "step": 8287
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.00808694557512e-06,
+      "loss": 1.1124,
+      "step": 8288
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.995362625834756e-06,
+      "loss": 1.2395,
+      "step": 8289
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.9826494494340756e-06,
+      "loss": 1.6205,
+      "step": 8290
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.969947417896338e-06,
+      "loss": 1.3047,
+      "step": 8291
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.957256532743417e-06,
+      "loss": 1.8182,
+      "step": 8292
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.944576795495905e-06,
+      "loss": 1.2601,
+      "step": 8293
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.931908207672977e-06,
+      "loss": 1.1762,
+      "step": 8294
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.919250770792574e-06,
+      "loss": 1.3329,
+      "step": 8295
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.906604486371204e-06,
+      "loss": 1.2778,
+      "step": 8296
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.893969355924068e-06,
+      "loss": 1.3629,
+      "step": 8297
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.881345380965088e-06,
+      "loss": 0.8805,
+      "step": 8298
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.868732563006763e-06,
+      "loss": 1.2304,
+      "step": 8299
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.856130903560331e-06,
+      "loss": 1.4782,
+      "step": 8300
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.843540404135629e-06,
+      "loss": 1.4737,
+      "step": 8301
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.830961066241226e-06,
+      "loss": 1.1897,
+      "step": 8302
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.818392891384262e-06,
+      "loss": 1.1851,
+      "step": 8303
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.8058358810706215e-06,
+      "loss": 1.3004,
+      "step": 8304
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.793290036804811e-06,
+      "loss": 1.1119,
+      "step": 8305
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.78075536009003e-06,
+      "loss": 1.259,
+      "step": 8306
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.768231852428075e-06,
+      "loss": 1.2494,
+      "step": 8307
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.755719515319503e-06,
+      "loss": 1.2867,
+      "step": 8308
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.7432183502634475e-06,
+      "loss": 1.2491,
+      "step": 8309
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.730728358757743e-06,
+      "loss": 1.4175,
+      "step": 8310
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.71824954229886e-06,
+      "loss": 1.4454,
+      "step": 8311
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.705781902381969e-06,
+      "loss": 1.3607,
+      "step": 8312
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.693325440500853e-06,
+      "loss": 1.2498,
+      "step": 8313
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.680880158148017e-06,
+      "loss": 1.1973,
+      "step": 8314
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.66844605681457e-06,
+      "loss": 1.1478,
+      "step": 8315
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.656023137990297e-06,
+      "loss": 1.1425,
+      "step": 8316
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.643611403163641e-06,
+      "loss": 1.3081,
+      "step": 8317
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.631210853821745e-06,
+      "loss": 1.2305,
+      "step": 8318
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.618821491450356e-06,
+      "loss": 1.2229,
+      "step": 8319
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.6064433175339055e-06,
+      "loss": 1.2111,
+      "step": 8320
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.594076333555488e-06,
+      "loss": 1.0921,
+      "step": 8321
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.581720540996861e-06,
+      "loss": 1.2723,
+      "step": 8322
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.5693759413384185e-06,
+      "loss": 1.3634,
+      "step": 8323
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.557042536059232e-06,
+      "loss": 1.5543,
+      "step": 8324
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.544720326637011e-06,
+      "loss": 1.2607,
+      "step": 8325
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.53240931454816e-06,
+      "loss": 1.2302,
+      "step": 8326
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.5201095012677125e-06,
+      "loss": 1.2604,
+      "step": 8327
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.507820888269389e-06,
+      "loss": 1.3969,
+      "step": 8328
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.495543477025523e-06,
+      "loss": 1.1171,
+      "step": 8329
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.483277269007149e-06,
+      "loss": 1.3911,
+      "step": 8330
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.471022265683912e-06,
+      "loss": 1.2743,
+      "step": 8331
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.458778468524163e-06,
+      "loss": 1.1968,
+      "step": 8332
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.446545878994914e-06,
+      "loss": 1.276,
+      "step": 8333
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.434324498561761e-06,
+      "loss": 1.345,
+      "step": 8334
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.422114328689066e-06,
+      "loss": 1.2564,
+      "step": 8335
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.4099153708397475e-06,
+      "loss": 1.2809,
+      "step": 8336
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.397727626475436e-06,
+      "loss": 1.3603,
+      "step": 8337
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.3855510970563966e-06,
+      "loss": 1.4609,
+      "step": 8338
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.373385784041563e-06,
+      "loss": 1.3806,
+      "step": 8339
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.361231688888525e-06,
+      "loss": 1.3766,
+      "step": 8340
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.349088813053516e-06,
+      "loss": 1.2838,
+      "step": 8341
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.336957157991452e-06,
+      "loss": 1.3057,
+      "step": 8342
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.3248367251558694e-06,
+      "loss": 1.0355,
+      "step": 8343
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.3127275159989615e-06,
+      "loss": 1.2395,
+      "step": 8344
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.300629531971636e-06,
+      "loss": 1.2635,
+      "step": 8345
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.288542774523354e-06,
+      "loss": 1.0626,
+      "step": 8346
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.276467245102347e-06,
+      "loss": 1.4016,
+      "step": 8347
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.2644029451553896e-06,
+      "loss": 1.4013,
+      "step": 8348
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.252349876128028e-06,
+      "loss": 1.1699,
+      "step": 8349
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.240308039464327e-06,
+      "loss": 1.4207,
+      "step": 8350
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.228277436607133e-06,
+      "loss": 1.4198,
+      "step": 8351
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.216258068997849e-06,
+      "loss": 1.5054,
+      "step": 8352
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.204249938076623e-06,
+      "loss": 1.5274,
+      "step": 8353
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.192253045282159e-06,
+      "loss": 1.4546,
+      "step": 8354
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.180267392051897e-06,
+      "loss": 1.3673,
+      "step": 8355
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.168292979821888e-06,
+      "loss": 1.5307,
+      "step": 8356
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.156329810026851e-06,
+      "loss": 1.1737,
+      "step": 8357
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.1443778841001166e-06,
+      "loss": 1.168,
+      "step": 8358
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.132437203473751e-06,
+      "loss": 1.0482,
+      "step": 8359
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.120507769578377e-06,
+      "loss": 1.2858,
+      "step": 8360
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.1085895838433715e-06,
+      "loss": 1.1256,
+      "step": 8361
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.096682647696684e-06,
+      "loss": 1.2619,
+      "step": 8362
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.084786962564937e-06,
+      "loss": 1.1232,
+      "step": 8363
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.072902529873403e-06,
+      "loss": 1.1772,
+      "step": 8364
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.061029351046043e-06,
+      "loss": 1.1939,
+      "step": 8365
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.049167427505409e-06,
+      "loss": 1.2074,
+      "step": 8366
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.037316760672773e-06,
+      "loss": 1.311,
+      "step": 8367
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.025477351967979e-06,
+      "loss": 1.3657,
+      "step": 8368
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.013649202809602e-06,
+      "loss": 1.2008,
+      "step": 8369
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.001832314614819e-06,
+      "loss": 1.8523,
+      "step": 8370
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.990026688799444e-06,
+      "loss": 1.429,
+      "step": 8371
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.978232326777999e-06,
+      "loss": 1.1324,
+      "step": 8372
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.9664492299636e-06,
+      "loss": 1.079,
+      "step": 8373
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.9546773997680605e-06,
+      "loss": 1.3254,
+      "step": 8374
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.942916837601809e-06,
+      "loss": 0.9873,
+      "step": 8375
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.931167544873939e-06,
+      "loss": 1.2471,
+      "step": 8376
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.91942952299217e-06,
+      "loss": 1.1243,
+      "step": 8377
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.907702773362933e-06,
+      "loss": 1.4397,
+      "step": 8378
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.895987297391226e-06,
+      "loss": 1.4521,
+      "step": 8379
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.884283096480769e-06,
+      "loss": 1.2006,
+      "step": 8380
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.872590172033865e-06,
+      "loss": 1.5175,
+      "step": 8381
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.860908525451547e-06,
+      "loss": 1.4846,
+      "step": 8382
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.849238158133408e-06,
+      "loss": 1.432,
+      "step": 8383
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.837579071477761e-06,
+      "loss": 1.1581,
+      "step": 8384
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.825931266881512e-06,
+      "loss": 1.4532,
+      "step": 8385
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.814294745740256e-06,
+      "loss": 0.8406,
+      "step": 8386
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.802669509448211e-06,
+      "loss": 1.3508,
+      "step": 8387
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.791055559398284e-06,
+      "loss": 1.3783,
+      "step": 8388
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.779452896981974e-06,
+      "loss": 1.209,
+      "step": 8389
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.7678615235894466e-06,
+      "loss": 1.1007,
+      "step": 8390
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.756281440609523e-06,
+      "loss": 0.9761,
+      "step": 8391
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.7447126494296954e-06,
+      "loss": 1.2318,
+      "step": 8392
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.7331551514360314e-06,
+      "loss": 1.0214,
+      "step": 8393
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.721608948013346e-06,
+      "loss": 1.5892,
+      "step": 8394
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.710074040544999e-06,
+      "loss": 1.1977,
+      "step": 8395
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.698550430413074e-06,
+      "loss": 1.2744,
+      "step": 8396
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.687038118998256e-06,
+      "loss": 1.3024,
+      "step": 8397
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.675537107679907e-06,
+      "loss": 1.4397,
+      "step": 8398
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.664047397835981e-06,
+      "loss": 1.2565,
+      "step": 8399
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.652568990843166e-06,
+      "loss": 1.2473,
+      "step": 8400
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.641101888076705e-06,
+      "loss": 1.4124,
+      "step": 8401
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.629646090910568e-06,
+      "loss": 1.5701,
+      "step": 8402
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.6182016007172986e-06,
+      "loss": 1.0412,
+      "step": 8403
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.606768418868114e-06,
+      "loss": 1.3912,
+      "step": 8404
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.595346546732905e-06,
+      "loss": 1.0955,
+      "step": 8405
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.583935985680155e-06,
+      "loss": 0.9747,
+      "step": 8406
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.57253673707705e-06,
+      "loss": 1.0456,
+      "step": 8407
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.561148802289351e-06,
+      "loss": 1.227,
+      "step": 8408
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.5497721826815565e-06,
+      "loss": 1.0869,
+      "step": 8409
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.538406879616698e-06,
+      "loss": 1.4057,
+      "step": 8410
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.527052894456541e-06,
+      "loss": 1.1552,
+      "step": 8411
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.515710228561433e-06,
+      "loss": 1.0152,
+      "step": 8412
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.50437888329044e-06,
+      "loss": 1.195,
+      "step": 8413
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.493058860001177e-06,
+      "loss": 1.68,
+      "step": 8414
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.481750160049981e-06,
+      "loss": 1.3081,
+      "step": 8415
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.4704527847918015e-06,
+      "loss": 1.4344,
+      "step": 8416
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.459166735580235e-06,
+      "loss": 1.2103,
+      "step": 8417
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.447892013767475e-06,
+      "loss": 1.5144,
+      "step": 8418
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.436628620704464e-06,
+      "loss": 1.3033,
+      "step": 8419
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.4253765577406805e-06,
+      "loss": 1.2668,
+      "step": 8420
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.414135826224309e-06,
+      "loss": 1.3531,
+      "step": 8421
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.402906427502164e-06,
+      "loss": 1.1566,
+      "step": 8422
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.391688362919678e-06,
+      "loss": 1.4394,
+      "step": 8423
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.380481633820922e-06,
+      "loss": 1.5118,
+      "step": 8424
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.3692862415486766e-06,
+      "loss": 1.2402,
+      "step": 8425
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.35810218744427e-06,
+      "loss": 1.1742,
+      "step": 8426
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.346929472847761e-06,
+      "loss": 1.2884,
+      "step": 8427
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.335768099097771e-06,
+      "loss": 1.0481,
+      "step": 8428
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.32461806753164e-06,
+      "loss": 0.9984,
+      "step": 8429
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.313479379485264e-06,
+      "loss": 1.3993,
+      "step": 8430
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.302352036293257e-06,
+      "loss": 1.2413,
+      "step": 8431
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.291236039288805e-06,
+      "loss": 1.4651,
+      "step": 8432
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.2801313898037994e-06,
+      "loss": 1.1393,
+      "step": 8433
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.269038089168721e-06,
+      "loss": 1.529,
+      "step": 8434
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.25795613871275e-06,
+      "loss": 1.2399,
+      "step": 8435
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.246885539763624e-06,
+      "loss": 1.2007,
+      "step": 8436
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.235826293647805e-06,
+      "loss": 1.1655,
+      "step": 8437
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.224778401690311e-06,
+      "loss": 1.5983,
+      "step": 8438
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.213741865214883e-06,
+      "loss": 1.3591,
+      "step": 8439
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.2027166855438295e-06,
+      "loss": 0.8701,
+      "step": 8440
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.19170286399816e-06,
+      "loss": 1.214,
+      "step": 8441
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.180700401897487e-06,
+      "loss": 1.2506,
+      "step": 8442
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.169709300560066e-06,
+      "loss": 1.3741,
+      "step": 8443
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.1587295613028e-06,
+      "loss": 1.2985,
+      "step": 8444
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.147761185441202e-06,
+      "loss": 1.3599,
+      "step": 8445
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.136804174289489e-06,
+      "loss": 1.2035,
+      "step": 8446
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.125858529160421e-06,
+      "loss": 1.2835,
+      "step": 8447
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.114924251365494e-06,
+      "loss": 1.4133,
+      "step": 8448
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.104001342214781e-06,
+      "loss": 1.1005,
+      "step": 8449
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.093089803017015e-06,
+      "loss": 1.4393,
+      "step": 8450
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.082189635079526e-06,
+      "loss": 1.2614,
+      "step": 8451
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.0713008397083705e-06,
+      "loss": 1.1228,
+      "step": 8452
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.060423418208149e-06,
+      "loss": 1.1247,
+      "step": 8453
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.049557371882163e-06,
+      "loss": 1.1014,
+      "step": 8454
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.038702702032294e-06,
+      "loss": 1.2146,
+      "step": 8455
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.027859409959135e-06,
+      "loss": 1.3128,
+      "step": 8456
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.017027496961857e-06,
+      "loss": 1.5506,
+      "step": 8457
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.006206964338267e-06,
+      "loss": 1.2805,
+      "step": 8458
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.995397813384839e-06,
+      "loss": 1.1785,
+      "step": 8459
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.984600045396681e-06,
+      "loss": 1.3179,
+      "step": 8460
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.9738136616675035e-06,
+      "loss": 1.4622,
+      "step": 8461
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.963038663489694e-06,
+      "loss": 1.2547,
+      "step": 8462
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.952275052154254e-06,
+      "loss": 1.0622,
+      "step": 8463
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.941522828950829e-06,
+      "loss": 1.5999,
+      "step": 8464
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.930781995167677e-06,
+      "loss": 1.2239,
+      "step": 8465
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.920052552091725e-06,
+      "loss": 1.4142,
+      "step": 8466
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.9093345010085e-06,
+      "loss": 1.2473,
+      "step": 8467
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.89862784320223e-06,
+      "loss": 1.4545,
+      "step": 8468
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.8879325799556894e-06,
+      "loss": 1.2521,
+      "step": 8469
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.877248712550342e-06,
+      "loss": 1.2251,
+      "step": 8470
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.866576242266263e-06,
+      "loss": 1.2457,
+      "step": 8471
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.8559151703822085e-06,
+      "loss": 1.2245,
+      "step": 8472
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.845265498175477e-06,
+      "loss": 1.4359,
+      "step": 8473
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.834627226922117e-06,
+      "loss": 1.0948,
+      "step": 8474
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.8240003578967184e-06,
+      "loss": 1.182,
+      "step": 8475
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.813384892372541e-06,
+      "loss": 1.2651,
+      "step": 8476
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.80278083162149e-06,
+      "loss": 1.3222,
+      "step": 8477
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.79218817691407e-06,
+      "loss": 1.1031,
+      "step": 8478
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.781606929519456e-06,
+      "loss": 1.0842,
+      "step": 8479
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.771037090705421e-06,
+      "loss": 1.4308,
+      "step": 8480
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.76047866173841e-06,
+      "loss": 1.4644,
+      "step": 8481
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.749931643883465e-06,
+      "loss": 1.1429,
+      "step": 8482
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.739396038404286e-06,
+      "loss": 1.3912,
+      "step": 8483
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.728871846563176e-06,
+      "loss": 1.6307,
+      "step": 8484
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.718359069621114e-06,
+      "loss": 0.9998,
+      "step": 8485
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7078577088376704e-06,
+      "loss": 1.3624,
+      "step": 8486
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.697367765471072e-06,
+      "loss": 1.2371,
+      "step": 8487
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.686889240778158e-06,
+      "loss": 1.0415,
+      "step": 8488
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.676422136014436e-06,
+      "loss": 0.765,
+      "step": 8489
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.665966452434012e-06,
+      "loss": 1.3054,
+      "step": 8490
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.655522191289619e-06,
+      "loss": 1.651,
+      "step": 8491
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.645089353832632e-06,
+      "loss": 1.4272,
+      "step": 8492
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.634667941313087e-06,
+      "loss": 1.2503,
+      "step": 8493
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.624257954979594e-06,
+      "loss": 0.8078,
+      "step": 8494
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6138593960794564e-06,
+      "loss": 0.8638,
+      "step": 8495
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.603472265858555e-06,
+      "loss": 1.5539,
+      "step": 8496
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.593096565561428e-06,
+      "loss": 1.1662,
+      "step": 8497
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.582732296431225e-06,
+      "loss": 1.2833,
+      "step": 8498
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.572379459709763e-06,
+      "loss": 0.9937,
+      "step": 8499
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.562038056637441e-06,
+      "loss": 0.6841,
+      "step": 8500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5517080884533436e-06,
+      "loss": 1.5837,
+      "step": 8501
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.541389556395126e-06,
+      "loss": 1.5026,
+      "step": 8502
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.531082461699121e-06,
+      "loss": 1.183,
+      "step": 8503
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.520786805600275e-06,
+      "loss": 0.7733,
+      "step": 8504
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.510502589332133e-06,
+      "loss": 1.0694,
+      "step": 8505
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.500229814126922e-06,
+      "loss": 1.4633,
+      "step": 8506
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.489968481215467e-06,
+      "loss": 1.1497,
+      "step": 8507
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.47971859182722e-06,
+      "loss": 1.3045,
+      "step": 8508
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.469480147190286e-06,
+      "loss": 1.2118,
+      "step": 8509
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.459253148531372e-06,
+      "loss": 1.3979,
+      "step": 8510
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.449037597075833e-06,
+      "loss": 1.2256,
+      "step": 8511
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.438833494047623e-06,
+      "loss": 1.1541,
+      "step": 8512
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.428640840669374e-06,
+      "loss": 1.1525,
+      "step": 8513
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.418459638162298e-06,
+      "loss": 1.4167,
+      "step": 8514
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.408289887746264e-06,
+      "loss": 1.806,
+      "step": 8515
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.398131590639765e-06,
+      "loss": 1.4672,
+      "step": 8516
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.387984748059903e-06,
+      "loss": 1.3454,
+      "step": 8517
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3778493612224305e-06,
+      "loss": 1.1234,
+      "step": 8518
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.367725431341696e-06,
+      "loss": 1.3674,
+      "step": 8519
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.35761295963073e-06,
+      "loss": 1.4298,
+      "step": 8520
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.347511947301119e-06,
+      "loss": 1.3465,
+      "step": 8521
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.337422395563162e-06,
+      "loss": 1.3952,
+      "step": 8522
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3273443056257025e-06,
+      "loss": 1.2223,
+      "step": 8523
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3172776786962635e-06,
+      "loss": 1.2094,
+      "step": 8524
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.307222515980958e-06,
+      "loss": 1.3044,
+      "step": 8525
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.297178818684555e-06,
+      "loss": 1.2084,
+      "step": 8526
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.287146588010438e-06,
+      "loss": 1.4106,
+      "step": 8527
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.277125825160621e-06,
+      "loss": 1.068,
+      "step": 8528
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.267116531335735e-06,
+      "loss": 1.265,
+      "step": 8529
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.257118707735053e-06,
+      "loss": 1.596,
+      "step": 8530
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.247132355556438e-06,
+      "loss": 1.207,
+      "step": 8531
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.237157475996434e-06,
+      "loss": 1.3766,
+      "step": 8532
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2271940702501625e-06,
+      "loss": 1.214,
+      "step": 8533
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.217242139511391e-06,
+      "loss": 1.0925,
+      "step": 8534
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.207301684972509e-06,
+      "loss": 1.6839,
+      "step": 8535
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1973727078245315e-06,
+      "loss": 1.3238,
+      "step": 8536
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.187455209257107e-06,
+      "loss": 1.0676,
+      "step": 8537
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1775491904584855e-06,
+      "loss": 1.0744,
+      "step": 8538
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.167654652615549e-06,
+      "loss": 1.0845,
+      "step": 8539
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.157771596913839e-06,
+      "loss": 1.1615,
+      "step": 8540
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.147900024537466e-06,
+      "loss": 1.1652,
+      "step": 8541
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.138039936669203e-06,
+      "loss": 1.6063,
+      "step": 8542
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.128191334490439e-06,
+      "loss": 0.8942,
+      "step": 8543
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1183542191811646e-06,
+      "loss": 1.5578,
+      "step": 8544
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.108528591920024e-06,
+      "loss": 1.1168,
+      "step": 8545
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.098714453884289e-06,
+      "loss": 1.5145,
+      "step": 8546
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.088911806249807e-06,
+      "loss": 1.2139,
+      "step": 8547
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.079120650191104e-06,
+      "loss": 1.4451,
+      "step": 8548
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.069340986881309e-06,
+      "loss": 1.2696,
+      "step": 8549
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0595728174921634e-06,
+      "loss": 1.1947,
+      "step": 8550
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.04981614319404e-06,
+      "loss": 1.5715,
+      "step": 8551
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.040070965155918e-06,
+      "loss": 1.4205,
+      "step": 8552
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.030337284545449e-06,
+      "loss": 1.4178,
+      "step": 8553
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.020615102528835e-06,
+      "loss": 1.3065,
+      "step": 8554
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.010904420270978e-06,
+      "loss": 1.3261,
+      "step": 8555
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.001205238935335e-06,
+      "loss": 1.3001,
+      "step": 8556
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.99151755968401e-06,
+      "loss": 1.4142,
+      "step": 8557
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.981841383677742e-06,
+      "loss": 0.8678,
+      "step": 8558
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9721767120758815e-06,
+      "loss": 1.2054,
+      "step": 8559
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.962523546036379e-06,
+      "loss": 1.1377,
+      "step": 8560
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.952881886715864e-06,
+      "loss": 1.2174,
+      "step": 8561
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.943251735269527e-06,
+      "loss": 1.1527,
+      "step": 8562
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.933633092851208e-06,
+      "loss": 1.3645,
+      "step": 8563
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.924025960613376e-06,
+      "loss": 1.2117,
+      "step": 8564
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.914430339707087e-06,
+      "loss": 1.0874,
+      "step": 8565
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.904846231282033e-06,
+      "loss": 1.4109,
+      "step": 8566
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.895273636486574e-06,
+      "loss": 1.278,
+      "step": 8567
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.885712556467602e-06,
+      "loss": 1.376,
+      "step": 8568
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.876162992370713e-06,
+      "loss": 1.2971,
+      "step": 8569
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.866624945340069e-06,
+      "loss": 1.5769,
+      "step": 8570
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.857098416518479e-06,
+      "loss": 1.543,
+      "step": 8571
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.84758340704734e-06,
+      "loss": 1.5278,
+      "step": 8572
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.838079918066717e-06,
+      "loss": 0.913,
+      "step": 8573
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.828587950715256e-06,
+      "loss": 1.149,
+      "step": 8574
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.819107506130249e-06,
+      "loss": 1.1147,
+      "step": 8575
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8096385854475837e-06,
+      "loss": 1.3702,
+      "step": 8576
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.800181189801799e-06,
+      "loss": 1.16,
+      "step": 8577
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.790735320325989e-06,
+      "loss": 1.1834,
+      "step": 8578
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.781300978151947e-06,
+      "loss": 0.9437,
+      "step": 8579
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.771878164410014e-06,
+      "loss": 1.2056,
+      "step": 8580
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.762466880229232e-06,
+      "loss": 1.204,
+      "step": 8581
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.753067126737153e-06,
+      "loss": 1.1677,
+      "step": 8582
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7436789050600664e-06,
+      "loss": 1.3372,
+      "step": 8583
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7343022163227826e-06,
+      "loss": 1.2872,
+      "step": 8584
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7249370616487813e-06,
+      "loss": 1.4137,
+      "step": 8585
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.715583442160142e-06,
+      "loss": 1.8864,
+      "step": 8586
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7062413589775693e-06,
+      "loss": 1.148,
+      "step": 8587
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.696910813220389e-06,
+      "loss": 1.4618,
+      "step": 8588
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6875918060065406e-06,
+      "loss": 1.2017,
+      "step": 8589
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6782843384525758e-06,
+      "loss": 1.5722,
+      "step": 8590
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.668988411673657e-06,
+      "loss": 1.0991,
+      "step": 8591
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6597040267835947e-06,
+      "loss": 1.3266,
+      "step": 8592
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.650431184894776e-06,
+      "loss": 1.2261,
+      "step": 8593
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6411698871182455e-06,
+      "loss": 1.273,
+      "step": 8594
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6319201345636156e-06,
+      "loss": 1.1386,
+      "step": 8595
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6226819283391776e-06,
+      "loss": 1.2659,
+      "step": 8596
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.613455269551802e-06,
+      "loss": 1.4399,
+      "step": 8597
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6042401593069595e-06,
+      "loss": 1.3783,
+      "step": 8598
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.5950365987087564e-06,
+      "loss": 1.2407,
+      "step": 8599
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.5858445888599434e-06,
+      "loss": 1.1341,
+      "step": 8600
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.5766641308618286e-06,
+      "loss": 1.3836,
+      "step": 8601
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.567495225814399e-06,
+      "loss": 1.1258,
+      "step": 8602
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.558337874816209e-06,
+      "loss": 1.1283,
+      "step": 8603
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.549192078964447e-06,
+      "loss": 1.2296,
+      "step": 8604
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.5400578393548935e-06,
+      "loss": 1.4562,
+      "step": 8605
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.530935157082016e-06,
+      "loss": 1.0607,
+      "step": 8606
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.5218240332387965e-06,
+      "loss": 1.1538,
+      "step": 8607
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.512724468916917e-06,
+      "loss": 1.3585,
+      "step": 8608
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.5036364652066166e-06,
+      "loss": 1.4261,
+      "step": 8609
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4945600231968025e-06,
+      "loss": 1.2537,
+      "step": 8610
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4854951439749593e-06,
+      "loss": 1.1501,
+      "step": 8611
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.476441828627175e-06,
+      "loss": 1.2691,
+      "step": 8612
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.467400078238181e-06,
+      "loss": 1.351,
+      "step": 8613
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4583698938913223e-06,
+      "loss": 1.5148,
+      "step": 8614
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4493512766685223e-06,
+      "loss": 1.0994,
+      "step": 8615
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4403442276503938e-06,
+      "loss": 1.3819,
+      "step": 8616
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4313487479160746e-06,
+      "loss": 1.5842,
+      "step": 8617
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4223648385433794e-06,
+      "loss": 1.1109,
+      "step": 8618
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.413392500608692e-06,
+      "loss": 1.1669,
+      "step": 8619
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4044317351870414e-06,
+      "loss": 1.3822,
+      "step": 8620
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3954825433520686e-06,
+      "loss": 1.1431,
+      "step": 8621
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3865449261760162e-06,
+      "loss": 0.9357,
+      "step": 8622
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3776188847297498e-06,
+      "loss": 1.419,
+      "step": 8623
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3687044200827356e-06,
+      "loss": 1.1004,
+      "step": 8624
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.359801533303064e-06,
+      "loss": 1.2901,
+      "step": 8625
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3509102254574153e-06,
+      "loss": 1.6831,
+      "step": 8626
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.342030497611115e-06,
+      "loss": 1.1462,
+      "step": 8627
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3331623508280896e-06,
+      "loss": 0.9354,
+      "step": 8628
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.324305786170867e-06,
+      "loss": 1.3719,
+      "step": 8629
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3154608047006097e-06,
+      "loss": 1.5473,
+      "step": 8630
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3066274074770585e-06,
+      "loss": 1.2941,
+      "step": 8631
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2978055955585895e-06,
+      "loss": 1.3411,
+      "step": 8632
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2889953700022016e-06,
+      "loss": 1.0239,
+      "step": 8633
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.280196731863472e-06,
+      "loss": 1.3857,
+      "step": 8634
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2714096821966246e-06,
+      "loss": 1.0137,
+      "step": 8635
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2626342220544614e-06,
+      "loss": 1.162,
+      "step": 8636
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.253870352488453e-06,
+      "loss": 1.0933,
+      "step": 8637
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.245118074548592e-06,
+      "loss": 0.9701,
+      "step": 8638
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2363773892835516e-06,
+      "loss": 1.4649,
+      "step": 8639
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.227648297740593e-06,
+      "loss": 1.3343,
+      "step": 8640
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.218930800965603e-06,
+      "loss": 1.1689,
+      "step": 8641
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.210224900003056e-06,
+      "loss": 1.3992,
+      "step": 8642
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.201530595896052e-06,
+      "loss": 1.0781,
+      "step": 8643
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1928478896863014e-06,
+      "loss": 1.3286,
+      "step": 8644
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1841767824141167e-06,
+      "loss": 1.4049,
+      "step": 8645
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1755172751184114e-06,
+      "loss": 1.4931,
+      "step": 8646
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.166869368836756e-06,
+      "loss": 1.3054,
+      "step": 8647
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.158233064605276e-06,
+      "loss": 1.2722,
+      "step": 8648
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.149608363458734e-06,
+      "loss": 0.9089,
+      "step": 8649
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.140995266430502e-06,
+      "loss": 1.4982,
+      "step": 8650
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1323937745525555e-06,
+      "loss": 1.1159,
+      "step": 8651
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.12380388885547e-06,
+      "loss": 1.0473,
+      "step": 8652
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1152256103684553e-06,
+      "loss": 1.5888,
+      "step": 8653
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1066589401193003e-06,
+      "loss": 1.1571,
+      "step": 8654
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0981038791344396e-06,
+      "loss": 1.2679,
+      "step": 8655
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.089560428438887e-06,
+      "loss": 1.205,
+      "step": 8656
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0810285890562786e-06,
+      "loss": 1.3,
+      "step": 8657
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0725083620088636e-06,
+      "loss": 0.9725,
+      "step": 8658
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.063999748317459e-06,
+      "loss": 1.1603,
+      "step": 8659
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0555027490015707e-06,
+      "loss": 1.3735,
+      "step": 8660
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0470173650792188e-06,
+      "loss": 1.1687,
+      "step": 8661
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0385435975671227e-06,
+      "loss": 1.0509,
+      "step": 8662
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.030081447480537e-06,
+      "loss": 1.2446,
+      "step": 8663
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0216309158333623e-06,
+      "loss": 1.3237,
+      "step": 8664
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.013192003638077e-06,
+      "loss": 1.2713,
+      "step": 8665
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0047647119058276e-06,
+      "loss": 1.3993,
+      "step": 8666
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.996349041646296e-06,
+      "loss": 1.31,
+      "step": 8667
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9879449938678304e-06,
+      "loss": 1.163,
+      "step": 8668
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9795525695773373e-06,
+      "loss": 1.2388,
+      "step": 8669
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9711717697803787e-06,
+      "loss": 1.4556,
+      "step": 8670
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9628025954810732e-06,
+      "loss": 1.0384,
+      "step": 8671
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.954445047682197e-06,
+      "loss": 1.4978,
+      "step": 8672
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9460991273850823e-06,
+      "loss": 1.7424,
+      "step": 8673
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.937764835589718e-06,
+      "loss": 1.253,
+      "step": 8674
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.929442173294661e-06,
+      "loss": 1.2472,
+      "step": 8675
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.921131141497102e-06,
+      "loss": 1.0027,
+      "step": 8676
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.912831741192823e-06,
+      "loss": 1.1389,
+      "step": 8677
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.904543973376206e-06,
+      "loss": 1.0197,
+      "step": 8678
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8962678390402565e-06,
+      "loss": 1.5322,
+      "step": 8679
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8880033391765815e-06,
+      "loss": 1.4697,
+      "step": 8680
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.879750474775378e-06,
+      "loss": 1.2525,
+      "step": 8681
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8715092468254766e-06,
+      "loss": 1.407,
+      "step": 8682
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8632796563142883e-06,
+      "loss": 1.3634,
+      "step": 8683
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8550617042278684e-06,
+      "loss": 1.1228,
+      "step": 8684
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8468553915508068e-06,
+      "loss": 1.1608,
+      "step": 8685
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8386607192663727e-06,
+      "loss": 1.4059,
+      "step": 8686
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8304776883563923e-06,
+      "loss": 1.3753,
+      "step": 8687
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.822306299801325e-06,
+      "loss": 1.4908,
+      "step": 8688
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8141465545802214e-06,
+      "loss": 1.5221,
+      "step": 8689
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.805998453670744e-06,
+      "loss": 1.4034,
+      "step": 8690
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7978619980491562e-06,
+      "loss": 1.0067,
+      "step": 8691
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.789737188690311e-06,
+      "loss": 1.207,
+      "step": 8692
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7816240265677085e-06,
+      "loss": 1.4762,
+      "step": 8693
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.773522512653404e-06,
+      "loss": 1.2534,
+      "step": 8694
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.765432647918087e-06,
+      "loss": 1.0294,
+      "step": 8695
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7573544333310385e-06,
+      "loss": 1.0887,
+      "step": 8696
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.749287869860173e-06,
+      "loss": 0.5182,
+      "step": 8697
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.741232958471951e-06,
+      "loss": 1.4464,
+      "step": 8698
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7331897001315e-06,
+      "loss": 1.1508,
+      "step": 8699
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7251580958024937e-06,
+      "loss": 1.2594,
+      "step": 8700
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.717138146447251e-06,
+      "loss": 1.2576,
+      "step": 8701
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7091298530266818e-06,
+      "loss": 1.0863,
+      "step": 8702
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.701133216500318e-06,
+      "loss": 1.4502,
+      "step": 8703
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6931482378262485e-06,
+      "loss": 1.5468,
+      "step": 8704
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6851749179611974e-06,
+      "loss": 1.2432,
+      "step": 8705
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.677213257860489e-06,
+      "loss": 1.0012,
+      "step": 8706
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6692632584780496e-06,
+      "loss": 1.2597,
+      "step": 8707
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.661324920766406e-06,
+      "loss": 1.5121,
+      "step": 8708
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.653398245676697e-06,
+      "loss": 1.2724,
+      "step": 8709
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.645483234158663e-06,
+      "loss": 1.2176,
+      "step": 8710
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6375798871606126e-06,
+      "loss": 1.2459,
+      "step": 8711
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.629688205629499e-06,
+      "loss": 1.1468,
+      "step": 8712
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6218081905108662e-06,
+      "loss": 1.0164,
+      "step": 8713
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.613939842748858e-06,
+      "loss": 1.4531,
+      "step": 8714
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6060831632862217e-06,
+      "loss": 1.3464,
+      "step": 8715
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5982381530642807e-06,
+      "loss": 1.1682,
+      "step": 8716
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5904048130230175e-06,
+      "loss": 1.2141,
+      "step": 8717
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5825831441009696e-06,
+      "loss": 1.692,
+      "step": 8718
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5747731472352875e-06,
+      "loss": 1.3787,
+      "step": 8719
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.566974823361712e-06,
+      "loss": 1.3354,
+      "step": 8720
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5591881734146283e-06,
+      "loss": 1.4155,
+      "step": 8721
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5514131983269573e-06,
+      "loss": 0.6496,
+      "step": 8722
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.543649899030287e-06,
+      "loss": 1.2583,
+      "step": 8723
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5358982764547735e-06,
+      "loss": 1.9311,
+      "step": 8724
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.528158331529162e-06,
+      "loss": 1.4609,
+      "step": 8725
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.520430065180801e-06,
+      "loss": 1.4358,
+      "step": 8726
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.512713478335693e-06,
+      "loss": 1.3427,
+      "step": 8727
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.505008571918366e-06,
+      "loss": 1.1006,
+      "step": 8728
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4973153468519807e-06,
+      "loss": 1.0778,
+      "step": 8729
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4896338040583335e-06,
+      "loss": 1.0367,
+      "step": 8730
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4819639444577656e-06,
+      "loss": 1.3522,
+      "step": 8731
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4743057689692318e-06,
+      "loss": 1.3963,
+      "step": 8732
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.466659278510308e-06,
+      "loss": 1.2508,
+      "step": 8733
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.459024473997151e-06,
+      "loss": 1.3185,
+      "step": 8734
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.451401356344529e-06,
+      "loss": 1.4877,
+      "step": 8735
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4437899264658117e-06,
+      "loss": 1.0176,
+      "step": 8736
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.436190185272946e-06,
+      "loss": 1.3333,
+      "step": 8737
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4286021336765052e-06,
+      "loss": 1.0707,
+      "step": 8738
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.421025772585639e-06,
+      "loss": 1.3193,
+      "step": 8739
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4134611029081324e-06,
+      "loss": 1.3048,
+      "step": 8740
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4059081255503044e-06,
+      "loss": 1.2362,
+      "step": 8741
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3983668414171545e-06,
+      "loss": 1.1944,
+      "step": 8742
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3908372514122257e-06,
+      "loss": 1.1235,
+      "step": 8743
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3833193564376852e-06,
+      "loss": 1.2421,
+      "step": 8744
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.375813157394258e-06,
+      "loss": 1.1942,
+      "step": 8745
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.368318655181323e-06,
+      "loss": 1.4476,
+      "step": 8746
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3608358506968296e-06,
+      "loss": 1.5786,
+      "step": 8747
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3533647448373274e-06,
+      "loss": 1.1344,
+      "step": 8748
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.345905338497967e-06,
+      "loss": 1.1674,
+      "step": 8749
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3384576325724996e-06,
+      "loss": 1.3323,
+      "step": 8750
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.331021627953267e-06,
+      "loss": 1.1443,
+      "step": 8751
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.323597325531224e-06,
+      "loss": 1.1245,
+      "step": 8752
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3161847261958804e-06,
+      "loss": 1.5647,
+      "step": 8753
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.308783830835415e-06,
+      "loss": 1.4082,
+      "step": 8754
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3013946403365293e-06,
+      "loss": 1.575,
+      "step": 8755
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2940171555845934e-06,
+      "loss": 1.3077,
+      "step": 8756
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2866513774635224e-06,
+      "loss": 1.2606,
+      "step": 8757
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.279297306855843e-06,
+      "loss": 1.2946,
+      "step": 8758
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2719549446426846e-06,
+      "loss": 1.1952,
+      "step": 8759
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2646242917037764e-06,
+      "loss": 1.1403,
+      "step": 8760
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.257305348917438e-06,
+      "loss": 1.2589,
+      "step": 8761
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.24999811716059e-06,
+      "loss": 1.4155,
+      "step": 8762
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2427025973087322e-06,
+      "loss": 1.2859,
+      "step": 8763
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.235418790235999e-06,
+      "loss": 1.292,
+      "step": 8764
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.228146696815081e-06,
+      "loss": 1.4414,
+      "step": 8765
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2208863179172922e-06,
+      "loss": 1.1776,
+      "step": 8766
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.213637654412537e-06,
+      "loss": 1.2495,
+      "step": 8767
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.206400707169298e-06,
+      "loss": 1.0113,
+      "step": 8768
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1991754770546823e-06,
+      "loss": 1.2269,
+      "step": 8769
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1919619649343858e-06,
+      "loss": 1.2406,
+      "step": 8770
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1847601716726838e-06,
+      "loss": 1.2297,
+      "step": 8771
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1775700981324534e-06,
+      "loss": 1.4353,
+      "step": 8772
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.170391745175182e-06,
+      "loss": 1.4636,
+      "step": 8773
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1632251136609384e-06,
+      "loss": 1.1868,
+      "step": 8774
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1560702044484018e-06,
+      "loss": 1.4678,
+      "step": 8775
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1489270183948197e-06,
+      "loss": 1.2598,
+      "step": 8776
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1417955563560634e-06,
+      "loss": 1.1264,
+      "step": 8777
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1346758191865934e-06,
+      "loss": 1.132,
+      "step": 8778
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.12756780773945e-06,
+      "loss": 1.5225,
+      "step": 8779
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1204715228662743e-06,
+      "loss": 1.1201,
+      "step": 8780
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1133869654173187e-06,
+      "loss": 1.0629,
+      "step": 8781
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1063141362414163e-06,
+      "loss": 1.0987,
+      "step": 8782
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0992530361859884e-06,
+      "loss": 1.0964,
+      "step": 8783
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.092203666097081e-06,
+      "loss": 1.2117,
+      "step": 8784
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.085166026819296e-06,
+      "loss": 1.4854,
+      "step": 8785
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.078140119195837e-06,
+      "loss": 1.126,
+      "step": 8786
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0711259440685305e-06,
+      "loss": 1.2968,
+      "step": 8787
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0641235022777707e-06,
+      "loss": 1.2173,
+      "step": 8788
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0571327946625753e-06,
+      "loss": 1.4402,
+      "step": 8789
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0501538220604966e-06,
+      "loss": 1.0991,
+      "step": 8790
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0431865853077657e-06,
+      "loss": 1.4139,
+      "step": 8791
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0362310852391154e-06,
+      "loss": 1.3758,
+      "step": 8792
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.029287322687945e-06,
+      "loss": 1.192,
+      "step": 8793
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0223552984862005e-06,
+      "loss": 1.28,
+      "step": 8794
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0154350134644727e-06,
+      "loss": 1.505,
+      "step": 8795
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.008526468451888e-06,
+      "loss": 1.3845,
+      "step": 8796
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0016296642762167e-06,
+      "loss": 1.3327,
+      "step": 8797
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.994744601763776e-06,
+      "loss": 1.4978,
+      "step": 8798
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9878712817395063e-06,
+      "loss": 1.3947,
+      "step": 8799
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.981009705026937e-06,
+      "loss": 1.3103,
+      "step": 8800
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.974159872448167e-06,
+      "loss": 1.4714,
+      "step": 8801
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.96732178482395e-06,
+      "loss": 1.4696,
+      "step": 8802
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9604954429735533e-06,
+      "loss": 1.2839,
+      "step": 8803
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9536808477149006e-06,
+      "loss": 1.4961,
+      "step": 8804
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9468779998644495e-06,
+      "loss": 0.9649,
+      "step": 8805
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9400869002373147e-06,
+      "loss": 1.482,
+      "step": 8806
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9333075496471564e-06,
+      "loss": 1.2958,
+      "step": 8807
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9265399489062365e-06,
+      "loss": 1.2657,
+      "step": 8808
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.919784098825428e-06,
+      "loss": 0.8473,
+      "step": 8809
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9130400002141723e-06,
+      "loss": 1.1894,
+      "step": 8810
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9063076538805236e-06,
+      "loss": 1.2274,
+      "step": 8811
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8995870606311027e-06,
+      "loss": 1.2695,
+      "step": 8812
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8928782212711327e-06,
+      "loss": 1.265,
+      "step": 8813
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.886181136604459e-06,
+      "loss": 1.1878,
+      "step": 8814
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8794958074334622e-06,
+      "loss": 1.1175,
+      "step": 8815
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8728222345591684e-06,
+      "loss": 1.0078,
+      "step": 8816
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8661604187811598e-06,
+      "loss": 1.0545,
+      "step": 8817
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8595103608976316e-06,
+      "loss": 1.4917,
+      "step": 8818
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8528720617053242e-06,
+      "loss": 1.1378,
+      "step": 8819
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8462455219996567e-06,
+      "loss": 1.4736,
+      "step": 8820
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8396307425745385e-06,
+      "loss": 1.2701,
+      "step": 8821
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.833027724222558e-06,
+      "loss": 1.4398,
+      "step": 8822
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8264364677348267e-06,
+      "loss": 1.1369,
+      "step": 8823
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8198569739010906e-06,
+      "loss": 1.3178,
+      "step": 8824
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8132892435096639e-06,
+      "loss": 1.1276,
+      "step": 8825
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.806733277347472e-06,
+      "loss": 1.4496,
+      "step": 8826
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8001890761999984e-06,
+      "loss": 1.2717,
+      "step": 8827
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.793656640851349e-06,
+      "loss": 1.3426,
+      "step": 8828
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7871359720841862e-06,
+      "loss": 1.1273,
+      "step": 8829
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.780627070679819e-06,
+      "loss": 1.3959,
+      "step": 8830
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7741299374180787e-06,
+      "loss": 1.5528,
+      "step": 8831
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7676445730774428e-06,
+      "loss": 1.2194,
+      "step": 8832
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7611709784349228e-06,
+      "loss": 1.268,
+      "step": 8833
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7547091542661876e-06,
+      "loss": 1.4768,
+      "step": 8834
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7482591013454285e-06,
+      "loss": 1.2535,
+      "step": 8835
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7418208204454833e-06,
+      "loss": 1.3456,
+      "step": 8836
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7353943123377347e-06,
+      "loss": 1.0807,
+      "step": 8837
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7289795777922003e-06,
+      "loss": 1.3337,
+      "step": 8838
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7225766175774317e-06,
+      "loss": 1.332,
+      "step": 8839
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.716185432460604e-06,
+      "loss": 1.198,
+      "step": 8840
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7098060232074941e-06,
+      "loss": 1.2119,
+      "step": 8841
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7034383905824237e-06,
+      "loss": 1.2123,
+      "step": 8842
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.69708253534836e-06,
+      "loss": 1.2756,
+      "step": 8843
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6907384582668051e-06,
+      "loss": 1.1661,
+      "step": 8844
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6844061600978844e-06,
+      "loss": 1.1869,
+      "step": 8845
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6780856416003022e-06,
+      "loss": 1.321,
+      "step": 8846
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6717769035313412e-06,
+      "loss": 1.1681,
+      "step": 8847
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6654799466468861e-06,
+      "loss": 1.0971,
+      "step": 8848
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6591947717014222e-06,
+      "loss": 1.0165,
+      "step": 8849
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6529213794479803e-06,
+      "loss": 1.311,
+      "step": 8850
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.646659770638237e-06,
+      "loss": 1.3634,
+      "step": 8851
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.640409946022392e-06,
+      "loss": 1.2953,
+      "step": 8852
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.634171906349291e-06,
+      "loss": 0.9195,
+      "step": 8853
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6279456523663362e-06,
+      "loss": 1.3435,
+      "step": 8854
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6217311848195305e-06,
+      "loss": 1.1333,
+      "step": 8855
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6155285044534452e-06,
+      "loss": 1.3031,
+      "step": 8856
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6093376120112747e-06,
+      "loss": 1.3838,
+      "step": 8857
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6031585082347588e-06,
+      "loss": 1.1706,
+      "step": 8858
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5969911938642722e-06,
+      "loss": 1.4189,
+      "step": 8859
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5908356696387128e-06,
+      "loss": 1.398,
+      "step": 8860
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5846919362956459e-06,
+      "loss": 1.3259,
+      "step": 8861
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5785599945711494e-06,
+      "loss": 1.502,
+      "step": 8862
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5724398451999468e-06,
+      "loss": 1.0288,
+      "step": 8863
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.566331488915307e-06,
+      "loss": 1.2267,
+      "step": 8864
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5602349264491112e-06,
+      "loss": 1.4226,
+      "step": 8865
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5541501585317975e-06,
+      "loss": 1.2188,
+      "step": 8866
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5480771858924381e-06,
+      "loss": 0.9816,
+      "step": 8867
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.542016009258651e-06,
+      "loss": 1.6205,
+      "step": 8868
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5359666293566555e-06,
+      "loss": 1.2625,
+      "step": 8869
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5299290469112604e-06,
+      "loss": 1.2248,
+      "step": 8870
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5239032626458649e-06,
+      "loss": 1.3767,
+      "step": 8871
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.517889277282447e-06,
+      "loss": 1.2607,
+      "step": 8872
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5118870915415528e-06,
+      "loss": 1.4074,
+      "step": 8873
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5058967061423513e-06,
+      "loss": 1.0267,
+      "step": 8874
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4999181218025682e-06,
+      "loss": 1.6542,
+      "step": 8875
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4939513392385418e-06,
+      "loss": 1.2792,
+      "step": 8876
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.487996359165178e-06,
+      "loss": 1.5594,
+      "step": 8877
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4820531822959615e-06,
+      "loss": 1.421,
+      "step": 8878
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4761218093429784e-06,
+      "loss": 1.3659,
+      "step": 8879
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4702022410168935e-06,
+      "loss": 1.2483,
+      "step": 8880
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4642944780269619e-06,
+      "loss": 1.3012,
+      "step": 8881
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.458398521081028e-06,
+      "loss": 1.2367,
+      "step": 8882
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4525143708855049e-06,
+      "loss": 0.8613,
+      "step": 8883
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4466420281454064e-06,
+      "loss": 1.2021,
+      "step": 8884
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4407814935643248e-06,
+      "loss": 1.4817,
+      "step": 8885
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4349327678444435e-06,
+      "loss": 1.4562,
+      "step": 8886
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4290958516865127e-06,
+      "loss": 1.5465,
+      "step": 8887
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4232707457898953e-06,
+      "loss": 1.2628,
+      "step": 8888
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4174574508525217e-06,
+      "loss": 0.9881,
+      "step": 8889
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.411655967570924e-06,
+      "loss": 1.503,
+      "step": 8890
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4058662966401903e-06,
+      "loss": 1.4033,
+      "step": 8891
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4000884387540102e-06,
+      "loss": 1.3613,
+      "step": 8892
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3943223946046524e-06,
+      "loss": 0.9809,
+      "step": 8893
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3885681648829974e-06,
+      "loss": 1.3145,
+      "step": 8894
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3828257502784603e-06,
+      "loss": 1.3127,
+      "step": 8895
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.377095151479091e-06,
+      "loss": 1.4334,
+      "step": 8896
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3713763691714732e-06,
+      "loss": 1.2506,
+      "step": 8897
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.365669404040848e-06,
+      "loss": 1.1859,
+      "step": 8898
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.359974256770935e-06,
+      "loss": 1.1138,
+      "step": 8899
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.354290928044144e-06,
+      "loss": 1.2769,
+      "step": 8900
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.348619418541408e-06,
+      "loss": 1.2009,
+      "step": 8901
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3429597289422502e-06,
+      "loss": 1.4307,
+      "step": 8902
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3373118599247946e-06,
+      "loss": 1.4258,
+      "step": 8903
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3316758121657335e-06,
+      "loss": 1.2481,
+      "step": 8904
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3260515863403599e-06,
+      "loss": 1.2942,
+      "step": 8905
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.320439183122535e-06,
+      "loss": 0.9782,
+      "step": 8906
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3148386031846981e-06,
+      "loss": 1.3677,
+      "step": 8907
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3092498471978908e-06,
+      "loss": 1.5745,
+      "step": 8908
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.303672915831744e-06,
+      "loss": 1.5227,
+      "step": 8909
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.298107809754423e-06,
+      "loss": 1.7287,
+      "step": 8910
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2925545296327502e-06,
+      "loss": 0.9877,
+      "step": 8911
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2870130761320488e-06,
+      "loss": 1.3222,
+      "step": 8912
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2814834499162986e-06,
+      "loss": 1.2021,
+      "step": 8913
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2759656516480145e-06,
+      "loss": 1.4023,
+      "step": 8914
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.270459681988334e-06,
+      "loss": 1.4456,
+      "step": 8915
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2649655415969185e-06,
+      "loss": 1.2073,
+      "step": 8916
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2594832311320747e-06,
+      "loss": 1.1105,
+      "step": 8917
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.254012751250666e-06,
+      "loss": 1.103,
+      "step": 8918
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2485541026081239e-06,
+      "loss": 1.5477,
+      "step": 8919
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2431072858584802e-06,
+      "loss": 1.2068,
+      "step": 8920
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2376723016543467e-06,
+      "loss": 1.3636,
+      "step": 8921
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2322491506469026e-06,
+      "loss": 1.1798,
+      "step": 8922
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2268378334859498e-06,
+      "loss": 1.0753,
+      "step": 8923
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2214383508198147e-06,
+      "loss": 1.2909,
+      "step": 8924
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.216050703295457e-06,
+      "loss": 1.377,
+      "step": 8925
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2106748915583832e-06,
+      "loss": 1.5598,
+      "step": 8926
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.205310916252711e-06,
+      "loss": 1.5624,
+      "step": 8927
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1999587780211042e-06,
+      "loss": 1.3691,
+      "step": 8928
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1946184775048496e-06,
+      "loss": 1.2232,
+      "step": 8929
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1892900153437693e-06,
+      "loss": 1.1905,
+      "step": 8930
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1839733921763297e-06,
+      "loss": 1.0627,
+      "step": 8931
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1786686086395104e-06,
+      "loss": 1.2569,
+      "step": 8932
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1733756653689032e-06,
+      "loss": 1.4881,
+      "step": 8933
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1680945629987006e-06,
+      "loss": 1.4671,
+      "step": 8934
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1628253021616408e-06,
+      "loss": 1.3022,
+      "step": 8935
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1575678834890635e-06,
+      "loss": 1.531,
+      "step": 8936
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.152322307610898e-06,
+      "loss": 1.035,
+      "step": 8937
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1470885751556304e-06,
+      "loss": 1.1345,
+      "step": 8938
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1418666867503368e-06,
+      "loss": 1.2309,
+      "step": 8939
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1366566430206837e-06,
+      "loss": 1.4178,
+      "step": 8940
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.131458444590916e-06,
+      "loss": 1.3986,
+      "step": 8941
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1262720920838354e-06,
+      "loss": 1.3465,
+      "step": 8942
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1210975861208783e-06,
+      "loss": 1.5578,
+      "step": 8943
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1159349273219932e-06,
+      "loss": 1.3676,
+      "step": 8944
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.110784116305763e-06,
+      "loss": 1.323,
+      "step": 8945
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1056451536893275e-06,
+      "loss": 0.9597,
+      "step": 8946
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1005180400884051e-06,
+      "loss": 1.1732,
+      "step": 8947
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0954027761173158e-06,
+      "loss": 1.7525,
+      "step": 8948
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0902993623889135e-06,
+      "loss": 1.2263,
+      "step": 8949
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0852077995146981e-06,
+      "loss": 1.5442,
+      "step": 8950
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0801280881047037e-06,
+      "loss": 1.1441,
+      "step": 8951
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0750602287675438e-06,
+      "loss": 1.439,
+      "step": 8952
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0700042221104323e-06,
+      "loss": 1.1258,
+      "step": 8953
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0649600687391514e-06,
+      "loss": 1.1656,
+      "step": 8954
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0599277692580734e-06,
+      "loss": 1.204,
+      "step": 8955
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0549073242701268e-06,
+      "loss": 1.2122,
+      "step": 8956
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.049898734376853e-06,
+      "loss": 1.396,
+      "step": 8957
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0449020001783493e-06,
+      "loss": 1.4574,
+      "step": 8958
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0399171222732817e-06,
+      "loss": 1.4826,
+      "step": 8959
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.03494410125895e-06,
+      "loss": 1.1976,
+      "step": 8960
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0299829377311554e-06,
+      "loss": 1.2079,
+      "step": 8961
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0250336322843445e-06,
+      "loss": 1.1986,
+      "step": 8962
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0200961855115099e-06,
+      "loss": 1.0688,
+      "step": 8963
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0151705980042448e-06,
+      "loss": 0.8879,
+      "step": 8964
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0102568703526881e-06,
+      "loss": 1.516,
+      "step": 8965
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0053550031455805e-06,
+      "loss": 1.2612,
+      "step": 8966
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0004649969702517e-06,
+      "loss": 1.3474,
+      "step": 8967
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.95586852412589e-07,
+      "loss": 1.3921,
+      "step": 8968
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.907205700570576e-07,
+      "loss": 1.1482,
+      "step": 8969
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.85866150486725e-07,
+      "loss": 1.0807,
+      "step": 8970
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.810235942832146e-07,
+      "loss": 1.548,
+      "step": 8971
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.761929020267401e-07,
+      "loss": 1.059,
+      "step": 8972
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.713740742960942e-07,
+      "loss": 1.6299,
+      "step": 8973
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.66567111668637e-07,
+      "loss": 0.9625,
+      "step": 8974
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.617720147203079e-07,
+      "loss": 1.1775,
+      "step": 8975
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.569887840256586e-07,
+      "loss": 1.0338,
+      "step": 8976
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.522174201577527e-07,
+      "loss": 1.1625,
+      "step": 8977
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.474579236882997e-07,
+      "loss": 1.4143,
+      "step": 8978
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.427102951875543e-07,
+      "loss": 1.1009,
+      "step": 8979
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.379745352243397e-07,
+      "loss": 1.2829,
+      "step": 8980
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.332506443660904e-07,
+      "loss": 1.1621,
+      "step": 8981
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.285386231787651e-07,
+      "loss": 1.3807,
+      "step": 8982
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.238384722269788e-07,
+      "loss": 1.0893,
+      "step": 8983
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.191501920738477e-07,
+      "loss": 1.3411,
+      "step": 8984
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.144737832811113e-07,
+      "loss": 1.384,
+      "step": 8985
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.098092464090546e-07,
+      "loss": 1.1381,
+      "step": 8986
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.051565820165752e-07,
+      "loss": 1.3898,
+      "step": 8987
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.005157906611383e-07,
+      "loss": 0.9105,
+      "step": 8988
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.958868728987657e-07,
+      "loss": 1.2893,
+      "step": 8989
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.912698292840693e-07,
+      "loss": 0.9049,
+      "step": 8990
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.866646603702511e-07,
+      "loss": 1.6868,
+      "step": 8991
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.820713667090697e-07,
+      "loss": 1.3054,
+      "step": 8992
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.774899488508736e-07,
+      "loss": 0.9211,
+      "step": 8993
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.729204073445906e-07,
+      "loss": 1.3289,
+      "step": 8994
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.68362742737705e-07,
+      "loss": 1.0991,
+      "step": 8995
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.63816955576302e-07,
+      "loss": 1.2851,
+      "step": 8996
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.592830464050349e-07,
+      "loss": 1.1996,
+      "step": 8997
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.547610157671359e-07,
+      "loss": 0.837,
+      "step": 8998
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.50250864204416e-07,
+      "loss": 1.5403,
+      "step": 8999
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.457525922572318e-07,
+      "loss": 1.0991,
+      "step": 9000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.412662004645744e-07,
+      "loss": 1.3187,
+      "step": 9001
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.367916893639693e-07,
+      "loss": 1.1394,
+      "step": 9002
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.323290594915212e-07,
+      "loss": 1.3823,
+      "step": 9003
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.278783113819244e-07,
+      "loss": 1.2396,
+      "step": 9004
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.234394455684636e-07,
+      "loss": 1.1226,
+      "step": 9005
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.190124625829465e-07,
+      "loss": 1.0409,
+      "step": 9006
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.145973629558268e-07,
+      "loss": 1.2947,
+      "step": 9007
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.101941472160701e-07,
+      "loss": 1.4764,
+      "step": 9008
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.058028158912545e-07,
+      "loss": 1.2484,
+      "step": 9009
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.014233695075368e-07,
+      "loss": 1.5357,
+      "step": 9010
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.970558085896418e-07,
+      "loss": 1.1458,
+      "step": 9011
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.927001336608509e-07,
+      "loss": 1.2962,
+      "step": 9012
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.883563452430576e-07,
+      "loss": 1.1745,
+      "step": 9013
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.840244438566902e-07,
+      "loss": 1.2617,
+      "step": 9014
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.797044300208001e-07,
+      "loss": 1.2495,
+      "step": 9015
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.753963042529732e-07,
+      "loss": 1.4419,
+      "step": 9016
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.711000670693857e-07,
+      "loss": 1.5073,
+      "step": 9017
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.668157189848036e-07,
+      "loss": 1.3239,
+      "step": 9018
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.625432605125493e-07,
+      "loss": 1.4699,
+      "step": 9019
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.582826921645247e-07,
+      "loss": 1.575,
+      "step": 9020
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.540340144511992e-07,
+      "loss": 1.3591,
+      "step": 9021
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.497972278816545e-07,
+      "loss": 1.2673,
+      "step": 9022
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.455723329634957e-07,
+      "loss": 1.3795,
+      "step": 9023
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.413593302029397e-07,
+      "loss": 1.4739,
+      "step": 9024
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.371582201047722e-07,
+      "loss": 1.4714,
+      "step": 9025
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.329690031723346e-07,
+      "loss": 1.3118,
+      "step": 9026
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.28791679907559e-07,
+      "loss": 1.757,
+      "step": 9027
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.246262508109669e-07,
+      "loss": 1.0419,
+      "step": 9028
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.20472716381615e-07,
+      "loss": 1.5917,
+      "step": 9029
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.163310771171828e-07,
+      "loss": 1.3605,
+      "step": 9030
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.122013335138844e-07,
+      "loss": 1.1551,
+      "step": 9031
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.080834860665352e-07,
+      "loss": 0.9961,
+      "step": 9032
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.039775352684963e-07,
+      "loss": 1.1092,
+      "step": 9033
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.998834816117516e-07,
+      "loss": 1.0477,
+      "step": 9034
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.958013255867867e-07,
+      "loss": 1.0743,
+      "step": 9035
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.917310676827548e-07,
+      "loss": 1.0899,
+      "step": 9036
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.87672708387288e-07,
+      "loss": 1.1938,
+      "step": 9037
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.836262481866751e-07,
+      "loss": 1.1593,
+      "step": 9038
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.795916875657172e-07,
+      "loss": 1.4315,
+      "step": 9039
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.755690270078275e-07,
+      "loss": 1.0342,
+      "step": 9040
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.715582669949649e-07,
+      "loss": 0.9993,
+      "step": 9041
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.675594080077008e-07,
+      "loss": 1.1843,
+      "step": 9042
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.635724505251406e-07,
+      "loss": 1.3305,
+      "step": 9043
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.595973950249912e-07,
+      "loss": 1.3891,
+      "step": 9044
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.55634241983527e-07,
+      "loss": 1.5066,
+      "step": 9045
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.516829918755796e-07,
+      "loss": 1.3838,
+      "step": 9046
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.477436451745589e-07,
+      "loss": 1.2464,
+      "step": 9047
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.438162023524875e-07,
+      "loss": 0.9816,
+      "step": 9048
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.39900663879911e-07,
+      "loss": 1.1907,
+      "step": 9049
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.359970302259766e-07,
+      "loss": 1.4751,
+      "step": 9050
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.321053018583878e-07,
+      "loss": 1.3729,
+      "step": 9051
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.282254792434494e-07,
+      "loss": 1.3938,
+      "step": 9052
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.243575628460007e-07,
+      "loss": 1.2664,
+      "step": 9053
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.205015531294822e-07,
+      "loss": 1.5499,
+      "step": 9054
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.166574505559241e-07,
+      "loss": 1.218,
+      "step": 9055
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.128252555858693e-07,
+      "loss": 1.0894,
+      "step": 9056
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.090049686785059e-07,
+      "loss": 1.2382,
+      "step": 9057
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.051965902915346e-07,
+      "loss": 1.1494,
+      "step": 9058
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.014001208812681e-07,
+      "loss": 1.5073,
+      "step": 9059
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.976155609025869e-07,
+      "loss": 1.3104,
+      "step": 9060
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.938429108089171e-07,
+      "loss": 0.9892,
+      "step": 9061
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.900821710522975e-07,
+      "loss": 0.9962,
+      "step": 9062
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.863333420833228e-07,
+      "loss": 1.1979,
+      "step": 9063
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.825964243511339e-07,
+      "loss": 1.4038,
+      "step": 9064
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.788714183034949e-07,
+      "loss": 1.2617,
+      "step": 9065
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.751583243867042e-07,
+      "loss": 1.3149,
+      "step": 9066
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.714571430456505e-07,
+      "loss": 1.3931,
+      "step": 9067
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.677678747237902e-07,
+      "loss": 1.3428,
+      "step": 9068
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.640905198631475e-07,
+      "loss": 1.1359,
+      "step": 9069
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.604250789043253e-07,
+      "loss": 1.459,
+      "step": 9070
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.567715522864946e-07,
+      "loss": 1.4162,
+      "step": 9071
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.531299404474277e-07,
+      "loss": 1.1844,
+      "step": 9072
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.495002438234087e-07,
+      "loss": 1.2025,
+      "step": 9073
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.458824628493453e-07,
+      "loss": 1.2359,
+      "step": 9074
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.422765979587019e-07,
+      "loss": 1.4164,
+      "step": 9075
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.386826495835107e-07,
+      "loss": 1.157,
+      "step": 9076
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.351006181543939e-07,
+      "loss": 0.7091,
+      "step": 9077
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.315305041005192e-07,
+      "loss": 0.9856,
+      "step": 9078
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.279723078496334e-07,
+      "loss": 1.3545,
+      "step": 9079
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.244260298280735e-07,
+      "loss": 1.3585,
+      "step": 9080
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.208916704607436e-07,
+      "loss": 1.3883,
+      "step": 9081
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.173692301710831e-07,
+      "loss": 1.3139,
+      "step": 9082
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.138587093811654e-07,
+      "loss": 1.2356,
+      "step": 9083
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.103601085115761e-07,
+      "loss": 1.3894,
+      "step": 9084
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.068734279815245e-07,
+      "loss": 1.5388,
+      "step": 9085
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.033986682087433e-07,
+      "loss": 1.1829,
+      "step": 9086
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.999358296095879e-07,
+      "loss": 1.3495,
+      "step": 9087
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.964849125989269e-07,
+      "loss": 0.8996,
+      "step": 9088
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.930459175902513e-07,
+      "loss": 1.1003,
+      "step": 9089
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.896188449955985e-07,
+      "loss": 1.5522,
+      "step": 9090
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.862036952255844e-07,
+      "loss": 1.1697,
+      "step": 9091
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.828004686893928e-07,
+      "loss": 1.3546,
+      "step": 9092
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.794091657947864e-07,
+      "loss": 1.3755,
+      "step": 9093
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.7602978694808453e-07,
+      "loss": 1.2883,
+      "step": 9094
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.726623325541857e-07,
+      "loss": 1.0814,
+      "step": 9095
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6930680301657814e-07,
+      "loss": 1.3002,
+      "step": 9096
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.659631987372959e-07,
+      "loss": 1.2411,
+      "step": 9097
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6263152011694064e-07,
+      "loss": 0.9874,
+      "step": 9098
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.5931176755471537e-07,
+      "loss": 1.3668,
+      "step": 9099
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.5600394144835747e-07,
+      "loss": 1.2714,
+      "step": 9100
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.5270804219420536e-07,
+      "loss": 1.2503,
+      "step": 9101
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.4942407018715437e-07,
+      "loss": 1.2289,
+      "step": 9102
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.461520258206675e-07,
+      "loss": 1.3253,
+      "step": 9103
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.4289190948679784e-07,
+      "loss": 1.1023,
+      "step": 9104
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.396437215761551e-07,
+      "loss": 0.9594,
+      "step": 9105
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.3640746247790576e-07,
+      "loss": 1.153,
+      "step": 9106
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.331831325798064e-07,
+      "loss": 1.1567,
+      "step": 9107
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.2997073226819227e-07,
+      "loss": 1.2784,
+      "step": 9108
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.2677026192793343e-07,
+      "loss": 1.2986,
+      "step": 9109
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.235817219425231e-07,
+      "loss": 1.4892,
+      "step": 9110
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.204051126939668e-07,
+      "loss": 1.5433,
+      "step": 9111
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.172404345628933e-07,
+      "loss": 0.9517,
+      "step": 9112
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.1408768792846606e-07,
+      "loss": 1.2352,
+      "step": 9113
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.109468731684385e-07,
+      "loss": 1.2316,
+      "step": 9114
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0781799065910955e-07,
+      "loss": 1.5305,
+      "step": 9115
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.047010407753904e-07,
+      "loss": 1.1888,
+      "step": 9116
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0159602389071573e-07,
+      "loss": 1.1494,
+      "step": 9117
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.985029403771212e-07,
+      "loss": 1.3539,
+      "step": 9118
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.9542179060522157e-07,
+      "loss": 1.0206,
+      "step": 9119
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.9235257494415476e-07,
+      "loss": 1.2477,
+      "step": 9120
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.892952937616712e-07,
+      "loss": 1.1205,
+      "step": 9121
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.8624994742408883e-07,
+      "loss": 1.2639,
+      "step": 9122
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.832165362962714e-07,
+      "loss": 1.69,
+      "step": 9123
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.8019506074167264e-07,
+      "loss": 1.315,
+      "step": 9124
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.77185521122303e-07,
+      "loss": 1.4241,
+      "step": 9125
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7418791779877394e-07,
+      "loss": 1.0199,
+      "step": 9126
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7120225113020935e-07,
+      "loss": 1.3532,
+      "step": 9127
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.682285214743564e-07,
+      "loss": 0.9275,
+      "step": 9128
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.6526672918751894e-07,
+      "loss": 1.0825,
+      "step": 9129
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.6231687462453536e-07,
+      "loss": 1.078,
+      "step": 9130
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.593789581388785e-07,
+      "loss": 1.3653,
+      "step": 9131
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.5645298008252224e-07,
+      "loss": 1.444,
+      "step": 9132
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.535389408060641e-07,
+      "loss": 1.1013,
+      "step": 9133
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.5063684065863577e-07,
+      "loss": 1.2071,
+      "step": 9134
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4774667998795916e-07,
+      "loss": 0.9156,
+      "step": 9135
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4486845914032394e-07,
+      "loss": 1.0676,
+      "step": 9136
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4200217846056536e-07,
+      "loss": 1.326,
+      "step": 9137
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.391478382921309e-07,
+      "loss": 1.8478,
+      "step": 9138
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.363054389769915e-07,
+      "loss": 1.6917,
+      "step": 9139
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.3347498085571916e-07,
+      "loss": 1.0842,
+      "step": 9140
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.3065646426744256e-07,
+      "loss": 1.3533,
+      "step": 9141
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.278498895498583e-07,
+      "loss": 1.329,
+      "step": 9142
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.2505525703925286e-07,
+      "loss": 1.3146,
+      "step": 9143
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.222725670704363e-07,
+      "loss": 0.8868,
+      "step": 9144
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.195018199768529e-07,
+      "loss": 0.6997,
+      "step": 9145
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.1674301609043723e-07,
+      "loss": 1.1869,
+      "step": 9146
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.139961557417692e-07,
+      "loss": 1.1829,
+      "step": 9147
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.1126123925995234e-07,
+      "loss": 1.2391,
+      "step": 9148
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0853826697265777e-07,
+      "loss": 1.3631,
+      "step": 9149
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0582723920615784e-07,
+      "loss": 1.284,
+      "step": 9150
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.031281562852595e-07,
+      "loss": 1.265,
+      "step": 9151
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.004410185333706e-07,
+      "loss": 1.3739,
+      "step": 9152
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.977658262724226e-07,
+      "loss": 1.6738,
+      "step": 9153
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.951025798229701e-07,
+      "loss": 0.9085,
+      "step": 9154
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.9245127950409126e-07,
+      "loss": 1.1698,
+      "step": 9155
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8981192563346526e-07,
+      "loss": 1.4098,
+      "step": 9156
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8718451852731696e-07,
+      "loss": 1.348,
+      "step": 9157
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8456905850045013e-07,
+      "loss": 1.5001,
+      "step": 9158
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8196554586623626e-07,
+      "loss": 1.3039,
+      "step": 9159
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.793739809366258e-07,
+      "loss": 1.4238,
+      "step": 9160
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.767943640221038e-07,
+      "loss": 1.4753,
+      "step": 9161
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.742266954317674e-07,
+      "loss": 1.3024,
+      "step": 9162
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.7167097547324825e-07,
+      "loss": 1.3858,
+      "step": 9163
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.6912720445276817e-07,
+      "loss": 1.2776,
+      "step": 9164
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.6659538267510553e-07,
+      "loss": 1.1611,
+      "step": 9165
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.6407551044361766e-07,
+      "loss": 1.7011,
+      "step": 9166
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.6156758806020754e-07,
+      "loss": 1.0119,
+      "step": 9167
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.590716158253681e-07,
+      "loss": 1.3158,
+      "step": 9168
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.565875940381601e-07,
+      "loss": 1.104,
+      "step": 9169
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.541155229962011e-07,
+      "loss": 1.3178,
+      "step": 9170
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5165540299567634e-07,
+      "loss": 0.9833,
+      "step": 9171
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4920723433136116e-07,
+      "loss": 1.1036,
+      "step": 9172
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4677101729656536e-07,
+      "loss": 1.1216,
+      "step": 9173
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.443467521831999e-07,
+      "loss": 1.0356,
+      "step": 9174
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4193443928171025e-07,
+      "loss": 1.5579,
+      "step": 9175
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3953407888114286e-07,
+      "loss": 1.2087,
+      "step": 9176
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3714567126907894e-07,
+      "loss": 1.2697,
+      "step": 9177
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.347692167317117e-07,
+      "loss": 1.442,
+      "step": 9178
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3240471555375787e-07,
+      "loss": 1.3283,
+      "step": 9179
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3005216801851304e-07,
+      "loss": 1.2621,
+      "step": 9180
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2771157440786284e-07,
+      "loss": 1.201,
+      "step": 9181
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.253829350022385e-07,
+      "loss": 1.2129,
+      "step": 9182
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2306625008063907e-07,
+      "loss": 1.391,
+      "step": 9183
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2076151992065364e-07,
+      "loss": 1.1199,
+      "step": 9184
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.1846874479841682e-07,
+      "loss": 0.9488,
+      "step": 9185
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.1618792498863117e-07,
+      "loss": 1.2344,
+      "step": 9186
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.1391906076457803e-07,
+      "loss": 1.3741,
+      "step": 9187
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.116621523980955e-07,
+      "loss": 1.4719,
+      "step": 9188
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.0941720015960063e-07,
+      "loss": 1.505,
+      "step": 9189
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.0718420431807827e-07,
+      "loss": 1.0642,
+      "step": 9190
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.0496316514105886e-07,
+      "loss": 1.1448,
+      "step": 9191
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.0275408289467391e-07,
+      "loss": 1.3831,
+      "step": 9192
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.0055695784357842e-07,
+      "loss": 1.539,
+      "step": 9193
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9837179025105067e-07,
+      "loss": 1.364,
+      "step": 9194
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9619858037888129e-07,
+      "loss": 1.3485,
+      "step": 9195
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9403732848747302e-07,
+      "loss": 1.1228,
+      "step": 9196
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9188803483576322e-07,
+      "loss": 1.1321,
+      "step": 9197
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8975069968127922e-07,
+      "loss": 1.0941,
+      "step": 9198
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8762532328009396e-07,
+      "loss": 1.3003,
+      "step": 9199
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.855119058868704e-07,
+      "loss": 1.3544,
+      "step": 9200
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8341044775481709e-07,
+      "loss": 1.2918,
+      "step": 9201
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.813209491357215e-07,
+      "loss": 1.2322,
+      "step": 9202
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7924341027993895e-07,
+      "loss": 1.0612,
+      "step": 9203
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7717783143639256e-07,
+      "loss": 1.0048,
+      "step": 9204
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7512421285257318e-07,
+      "loss": 1.407,
+      "step": 9205
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.730825547745285e-07,
+      "loss": 1.0557,
+      "step": 9206
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7105285744687394e-07,
+      "loss": 1.2855,
+      "step": 9207
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6903512111280384e-07,
+      "loss": 1.421,
+      "step": 9208
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6702934601408038e-07,
+      "loss": 1.1971,
+      "step": 9209
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6503553239100022e-07,
+      "loss": 1.4362,
+      "step": 9210
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6305368048248336e-07,
+      "loss": 1.3482,
+      "step": 9211
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6108379052597323e-07,
+      "loss": 1.0659,
+      "step": 9212
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5912586275749208e-07,
+      "loss": 1.2496,
+      "step": 9213
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5717989741160788e-07,
+      "loss": 1.4317,
+      "step": 9214
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5524589472151185e-07,
+      "loss": 1.4145,
+      "step": 9215
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5332385491890756e-07,
+      "loss": 1.0512,
+      "step": 9216
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5141377823407742e-07,
+      "loss": 1.2391,
+      "step": 9217
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4951566489589396e-07,
+      "loss": 1.3951,
+      "step": 9218
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4762951513176415e-07,
+      "loss": 1.2797,
+      "step": 9219
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4575532916768497e-07,
+      "loss": 1.691,
+      "step": 9220
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4389310722822124e-07,
+      "loss": 1.318,
+      "step": 9221
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.420428495364723e-07,
+      "loss": 1.466,
+      "step": 9222
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4020455631413853e-07,
+      "loss": 1.2957,
+      "step": 9223
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3837822778147712e-07,
+      "loss": 1.2411,
+      "step": 9224
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3656386415730195e-07,
+      "loss": 1.1202,
+      "step": 9225
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3476146565901682e-07,
+      "loss": 1.2342,
+      "step": 9226
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3297103250256015e-07,
+      "loss": 1.1467,
+      "step": 9227
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3119256490244924e-07,
+      "loss": 1.298,
+      "step": 9228
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.294260630717803e-07,
+      "loss": 1.2076,
+      "step": 9229
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2767152722220621e-07,
+      "loss": 1.3032,
+      "step": 9230
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2592895756394773e-07,
+      "loss": 1.1198,
+      "step": 9231
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.241983543057934e-07,
+      "loss": 1.2922,
+      "step": 9232
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2247971765508848e-07,
+      "loss": 1.5338,
+      "step": 9233
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2077304781774602e-07,
+      "loss": 1.0034,
+      "step": 9234
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1907834499825798e-07,
+      "loss": 1.2603,
+      "step": 9235
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1739560939967309e-07,
+      "loss": 1.3585,
+      "step": 9236
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1572484122361893e-07,
+      "loss": 1.2684,
+      "step": 9237
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1406604067025762e-07,
+      "loss": 1.7801,
+      "step": 9238
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1241920793835237e-07,
+      "loss": 1.2115,
+      "step": 9239
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1078434322522313e-07,
+      "loss": 1.3815,
+      "step": 9240
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0916144672673545e-07,
+      "loss": 1.6389,
+      "step": 9241
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0755051863734489e-07,
+      "loss": 1.0366,
+      "step": 9242
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0595155915006372e-07,
+      "loss": 1.6249,
+      "step": 9243
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0436456845647203e-07,
+      "loss": 1.4272,
+      "step": 9244
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.027895467467066e-07,
+      "loss": 1.1991,
+      "step": 9245
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0122649420949426e-07,
+      "loss": 1.214,
+      "step": 9246
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.967541103210743e-08,
+      "loss": 1.315,
+      "step": 9247
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.813629740037522e-08,
+      "loss": 1.4949,
+      "step": 9248
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.66091534987168e-08,
+      "loss": 1.2652,
+      "step": 9249
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.509397951011911e-08,
+      "loss": 1.2505,
+      "step": 9250
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.359077561609254e-08,
+      "loss": 1.1525,
+      "step": 9251
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.209954199677073e-08,
+      "loss": 1.2255,
+      "step": 9252
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.062027883081081e-08,
+      "loss": 1.4147,
+      "step": 9253
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.915298629545987e-08,
+      "loss": 1.1834,
+      "step": 9254
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.769766456651062e-08,
+      "loss": 1.4864,
+      "step": 9255
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.625431381834582e-08,
+      "loss": 1.1522,
+      "step": 9256
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.482293422388266e-08,
+      "loss": 1.2748,
+      "step": 9257
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.340352595463952e-08,
+      "loss": 1.4174,
+      "step": 9258
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.199608918066925e-08,
+      "loss": 1.5165,
+      "step": 9259
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.060062407061475e-08,
+      "loss": 1.4426,
+      "step": 9260
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.921713079166449e-08,
+      "loss": 1.1407,
+      "step": 9261
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.784560950958587e-08,
+      "loss": 1.4924,
+      "step": 9262
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.648606038870299e-08,
+      "loss": 1.5181,
+      "step": 9263
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.513848359191888e-08,
+      "loss": 1.3165,
+      "step": 9264
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.38028792806933e-08,
+      "loss": 0.896,
+      "step": 9265
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.247924761504266e-08,
+      "loss": 1.1867,
+      "step": 9266
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.116758875356233e-08,
+      "loss": 0.9451,
+      "step": 9267
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.986790285340438e-08,
+      "loss": 1.4154,
+      "step": 9268
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.858019007028871e-08,
+      "loss": 1.109,
+      "step": 9269
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.730445055851409e-08,
+      "loss": 1.4334,
+      "step": 9270
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.604068447092493e-08,
+      "loss": 1.4307,
+      "step": 9271
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.478889195894456e-08,
+      "loss": 1.251,
+      "step": 9272
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.354907317255298e-08,
+      "loss": 1.5551,
+      "step": 9273
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.232122826029807e-08,
+      "loss": 1.3481,
+      "step": 9274
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.110535736929546e-08,
+      "loss": 1.0623,
+      "step": 9275
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.990146064521752e-08,
+      "loss": 1.7288,
+      "step": 9276
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.870953823232661e-08,
+      "loss": 1.5316,
+      "step": 9277
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.752959027340854e-08,
+      "loss": 1.1307,
+      "step": 9278
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.63616169098613e-08,
+      "loss": 1.8188,
+      "step": 9279
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5205618281617414e-08,
+      "loss": 1.1548,
+      "step": 9280
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.406159452717719e-08,
+      "loss": 1.5085,
+      "step": 9281
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.2929545783608757e-08,
+      "loss": 1.1656,
+      "step": 9282
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.180947218655918e-08,
+      "loss": 1.2269,
+      "step": 9283
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.0701373870232215e-08,
+      "loss": 1.0343,
+      "step": 9284
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.9605250967377225e-08,
+      "loss": 1.5754,
+      "step": 9285
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.85211036093447e-08,
+      "loss": 1.406,
+      "step": 9286
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.744893192601962e-08,
+      "loss": 1.2681,
+      "step": 9287
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.638873604587701e-08,
+      "loss": 1.4261,
+      "step": 9288
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.5340516095926375e-08,
+      "loss": 1.4359,
+      "step": 9289
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.4304272201778354e-08,
+      "loss": 1.1884,
+      "step": 9290
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.3280004487578074e-08,
+      "loss": 1.2303,
+      "step": 9291
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.2267713076038495e-08,
+      "loss": 1.5134,
+      "step": 9292
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.1267398088473686e-08,
+      "loss": 1.2121,
+      "step": 9293
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0279059644721115e-08,
+      "loss": 1.2688,
+      "step": 9294
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.930269786319718e-08,
+      "loss": 1.1604,
+      "step": 9295
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.8338312860886074e-08,
+      "loss": 1.5923,
+      "step": 9296
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7385904753339805e-08,
+      "loss": 1.3127,
+      "step": 9297
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.644547365466711e-08,
+      "loss": 0.9667,
+      "step": 9298
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.5517019677555607e-08,
+      "loss": 1.2342,
+      "step": 9299
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4600542933227456e-08,
+      "loss": 1.0964,
+      "step": 9300
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.3696043531505905e-08,
+      "loss": 1.2812,
+      "step": 9301
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.280352158074873e-08,
+      "loss": 1.0195,
+      "step": 9302
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.192297718791482e-08,
+      "loss": 1.2099,
+      "step": 9303
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.105441045849755e-08,
+      "loss": 1.3496,
+      "step": 9304
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.019782149654704e-08,
+      "loss": 1.2986,
+      "step": 9305
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.9353210404725606e-08,
+      "loss": 1.5576,
+      "step": 9306
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.8520577284207872e-08,
+      "loss": 1.0743,
+      "step": 9307
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7699922234769582e-08,
+      "loss": 1.2976,
+      "step": 9308
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.6891245354720985e-08,
+      "loss": 1.4146,
+      "step": 9309
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.6094546740962344e-08,
+      "loss": 1.252,
+      "step": 9310
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.530982648896174e-08,
+      "loss": 1.1864,
+      "step": 9311
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.4537084692721756e-08,
+      "loss": 1.2134,
+      "step": 9312
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3776321444834993e-08,
+      "loss": 1.027,
+      "step": 9313
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.302753683646186e-08,
+      "loss": 1.3234,
+      "step": 9314
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.229073095729728e-08,
+      "loss": 1.6696,
+      "step": 9315
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.1565903895637284e-08,
+      "loss": 1.1691,
+      "step": 9316
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.0853055738323524e-08,
+      "loss": 1.3213,
+      "step": 9317
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.0152186570776553e-08,
+      "loss": 1.3915,
+      "step": 9318
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.946329647694034e-08,
+      "loss": 1.2309,
+      "step": 9319
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.878638553939327e-08,
+      "loss": 1.2684,
+      "step": 9320
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8121453839203828e-08,
+      "loss": 1.2743,
+      "step": 9321
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7468501456063823e-08,
+      "loss": 1.3031,
+      "step": 9322
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.682752846819957e-08,
+      "loss": 1.0069,
+      "step": 9323
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6198534952405198e-08,
+      "loss": 1.1959,
+      "step": 9324
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5581520984053744e-08,
+      "loss": 1.5175,
+      "step": 9325
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.4976486637063856e-08,
+      "loss": 1.4378,
+      "step": 9326
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.4383431983933104e-08,
+      "loss": 1.4284,
+      "step": 9327
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3802357095715757e-08,
+      "loss": 1.2507,
+      "step": 9328
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3233262042022798e-08,
+      "loss": 1.1549,
+      "step": 9329
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2676146891055229e-08,
+      "loss": 1.2793,
+      "step": 9330
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2131011709559659e-08,
+      "loss": 1.219,
+      "step": 9331
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1597856562850506e-08,
+      "loss": 1.144,
+      "step": 9332
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1076681514810005e-08,
+      "loss": 1.2033,
+      "step": 9333
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.05674866278771e-08,
+      "loss": 1.3266,
+      "step": 9334
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0070271963069644e-08,
+      "loss": 1.4408,
+      "step": 9335
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.585037579951106e-09,
+      "loss": 1.2639,
+      "step": 9336
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.111783536663865e-09,
+      "loss": 1.4986,
+      "step": 9337
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.650509889918113e-09,
+      "loss": 1.7304,
+      "step": 9338
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.20121669496965e-09,
+      "loss": 1.0316,
+      "step": 9339
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.763904005653189e-09,
+      "loss": 1.4083,
+      "step": 9340
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.338571874371258e-09,
+      "loss": 1.2976,
+      "step": 9341
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.925220352071993e-09,
+      "loss": 1.5486,
+      "step": 9342
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.523849488304645e-09,
+      "loss": 1.2638,
+      "step": 9343
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.134459331130771e-09,
+      "loss": 1.2112,
+      "step": 9344
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.757049927235248e-09,
+      "loss": 1.4581,
+      "step": 9345
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.391621321815255e-09,
+      "loss": 1.1104,
+      "step": 9346
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.038173558657988e-09,
+      "loss": 1.3579,
+      "step": 9347
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.6967066801184565e-09,
+      "loss": 1.0755,
+      "step": 9348
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.367220727097277e-09,
+      "loss": 1.2945,
+      "step": 9349
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.049715739085081e-09,
+      "loss": 1.1304,
+      "step": 9350
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.744191754129211e-09,
+      "loss": 1.4249,
+      "step": 9351
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4506488088115186e-09,
+      "loss": 1.3096,
+      "step": 9352
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.1690869383260713e-09,
+      "loss": 1.2095,
+      "step": 9353
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.899506176390343e-09,
+      "loss": 1.2363,
+      "step": 9354
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.641906555322926e-09,
+      "loss": 1.4832,
+      "step": 9355
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.396288105965816e-09,
+      "loss": 0.8361,
+      "step": 9356
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.1626508577732297e-09,
+      "loss": 1.3599,
+      "step": 9357
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9409948387227872e-09,
+      "loss": 1.6008,
+      "step": 9358
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.7313200753599212e-09,
+      "loss": 1.5372,
+      "step": 9359
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5336265928422856e-09,
+      "loss": 1.1596,
+      "step": 9360
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3479144148176303e-09,
+      "loss": 1.4738,
+      "step": 9361
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.174183563568132e-09,
+      "loss": 1.396,
+      "step": 9362
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0124340598882676e-09,
+      "loss": 1.4128,
+      "step": 9363
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.626659231625311e-10,
+      "loss": 1.1626,
+      "step": 9364
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.24879171343229e-10,
+      "loss": 1.0485,
+      "step": 9365
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.990738209393776e-10,
+      "loss": 1.3435,
+      "step": 9366
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.852498870167033e-10,
+      "loss": 1.3852,
+      "step": 9367
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.8340738321984705e-10,
+      "loss": 0.9418,
+      "step": 9368
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.9354632173905773e-10,
+      "loss": 1.4683,
+      "step": 9369
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.1566671335460088e-10,
+      "loss": 1.263,
+      "step": 9370
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.497685673923499e-10,
+      "loss": 1.0752,
+      "step": 9371
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.585189174599052e-11,
+      "loss": 1.2421,
+      "step": 9372
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.391669287702072e-11,
+      "loss": 1.3334,
+      "step": 9373
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.396297580364859e-11,
+      "loss": 1.3883,
+      "step": 9374
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.990744134098947e-12,
+      "loss": 1.2799,
+      "step": 9375
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0,
+      "loss": 1.5744,
+      "step": 9376
+    },
+    {
+      "epoch": 2.0,
+      "step": 9376,
+      "total_flos": 1.3162175477630632e+19,
+      "train_loss": 1.319023442537279,
+      "train_runtime": 50524.683,
+      "train_samples_per_second": 0.186,
+      "train_steps_per_second": 0.186
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 9376,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 1000,
+  "total_flos": 1.3162175477630632e+19,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}