diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,58636 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 9768,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0204081632653061e-07,
+      "loss": 1.5146,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0408163265306121e-07,
+      "loss": 1.5576,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0612244897959183e-07,
+      "loss": 1.2544,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0816326530612243e-07,
+      "loss": 1.4209,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.102040816326531e-07,
+      "loss": 1.4795,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.122448979591837e-07,
+      "loss": 1.5117,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.142857142857143e-07,
+      "loss": 1.5381,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.163265306122449e-07,
+      "loss": 1.5723,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.183673469387755e-07,
+      "loss": 1.4248,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0204081632653063e-06,
+      "loss": 1.2637,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1224489795918367e-06,
+      "loss": 1.3301,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2244897959183673e-06,
+      "loss": 1.3223,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.326530612244898e-06,
+      "loss": 1.3428,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4285714285714286e-06,
+      "loss": 1.3325,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5306122448979593e-06,
+      "loss": 1.043,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6326530612244897e-06,
+      "loss": 1.1445,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7346938775510206e-06,
+      "loss": 1.2065,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.836734693877551e-06,
+      "loss": 1.0381,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.938775510204082e-06,
+      "loss": 1.1694,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0408163265306125e-06,
+      "loss": 1.0698,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1428571428571427e-06,
+      "loss": 1.0581,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2448979591836734e-06,
+      "loss": 1.0469,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3469387755102044e-06,
+      "loss": 0.9985,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4489795918367347e-06,
+      "loss": 1.0391,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5510204081632653e-06,
+      "loss": 0.978,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.653061224489796e-06,
+      "loss": 1.0327,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7551020408163266e-06,
+      "loss": 1.0645,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 0.9116,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.959183673469388e-06,
+      "loss": 0.9097,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0612244897959185e-06,
+      "loss": 0.8706,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.163265306122449e-06,
+      "loss": 0.9692,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2653061224489794e-06,
+      "loss": 0.9653,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3673469387755105e-06,
+      "loss": 0.8413,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.469387755102041e-06,
+      "loss": 0.9531,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5714285714285714e-06,
+      "loss": 0.8179,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.673469387755102e-06,
+      "loss": 0.9087,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7755102040816327e-06,
+      "loss": 0.9458,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.877551020408164e-06,
+      "loss": 0.9214,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.979591836734694e-06,
+      "loss": 0.8159,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.081632653061225e-06,
+      "loss": 0.8418,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.183673469387755e-06,
+      "loss": 0.9302,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 0.8643,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3877551020408165e-06,
+      "loss": 0.8438,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.489795918367347e-06,
+      "loss": 0.8574,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.591836734693878e-06,
+      "loss": 0.8071,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.693877551020409e-06,
+      "loss": 0.7446,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.795918367346939e-06,
+      "loss": 0.7939,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.897959183673469e-06,
+      "loss": 0.7417,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9999999999999996e-06,
+      "loss": 0.7427,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.102040816326531e-06,
+      "loss": 0.7764,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.204081632653062e-06,
+      "loss": 0.814,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.306122448979592e-06,
+      "loss": 0.678,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.408163265306123e-06,
+      "loss": 0.7646,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.510204081632653e-06,
+      "loss": 0.7734,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.612244897959184e-06,
+      "loss": 0.7568,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 0.7241,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.816326530612245e-06,
+      "loss": 0.7781,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.918367346938776e-06,
+      "loss": 0.8237,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.020408163265306e-06,
+      "loss": 0.7964,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.122448979591837e-06,
+      "loss": 0.7871,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.224489795918368e-06,
+      "loss": 0.748,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.326530612244898e-06,
+      "loss": 0.7412,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.428571428571429e-06,
+      "loss": 0.751,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.530612244897959e-06,
+      "loss": 0.6921,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.63265306122449e-06,
+      "loss": 0.6873,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.734693877551021e-06,
+      "loss": 0.7324,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.836734693877551e-06,
+      "loss": 0.791,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.938775510204082e-06,
+      "loss": 0.7256,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.0408163265306125e-06,
+      "loss": 0.7202,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.142857142857143e-06,
+      "loss": 0.7319,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.244897959183674e-06,
+      "loss": 0.8096,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.346938775510204e-06,
+      "loss": 0.7527,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.448979591836735e-06,
+      "loss": 0.6194,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.551020408163265e-06,
+      "loss": 0.8223,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.653061224489796e-06,
+      "loss": 0.6255,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.755102040816327e-06,
+      "loss": 0.7693,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.857142857142858e-06,
+      "loss": 0.657,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.959183673469388e-06,
+      "loss": 0.6758,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.06122448979592e-06,
+      "loss": 0.6948,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.16326530612245e-06,
+      "loss": 0.791,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.265306122448979e-06,
+      "loss": 0.793,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.36734693877551e-06,
+      "loss": 0.7461,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.46938775510204e-06,
+      "loss": 0.728,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.6846,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.673469387755101e-06,
+      "loss": 0.687,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.775510204081633e-06,
+      "loss": 0.7241,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.877551020408163e-06,
+      "loss": 0.604,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.979591836734694e-06,
+      "loss": 0.7271,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.081632653061225e-06,
+      "loss": 0.739,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.183673469387756e-06,
+      "loss": 0.6758,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.285714285714286e-06,
+      "loss": 0.6919,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.387755102040818e-06,
+      "loss": 0.7759,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.489795918367348e-06,
+      "loss": 0.7014,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.591836734693878e-06,
+      "loss": 0.7148,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.693877551020408e-06,
+      "loss": 0.7524,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.795918367346939e-06,
+      "loss": 0.6338,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.897959183673469e-06,
+      "loss": 0.7417,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999999999999999e-06,
+      "loss": 0.582,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0102040816326531e-05,
+      "loss": 0.6646,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0204081632653061e-05,
+      "loss": 0.6621,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0306122448979591e-05,
+      "loss": 0.6255,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0408163265306123e-05,
+      "loss": 0.7061,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0510204081632654e-05,
+      "loss": 0.665,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0612244897959184e-05,
+      "loss": 0.6555,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 0.6626,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0816326530612246e-05,
+      "loss": 0.6333,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0918367346938776e-05,
+      "loss": 0.6904,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1020408163265306e-05,
+      "loss": 0.7122,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1122448979591838e-05,
+      "loss": 0.6899,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1224489795918369e-05,
+      "loss": 0.708,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1326530612244897e-05,
+      "loss": 0.6782,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 0.7119,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.153061224489796e-05,
+      "loss": 0.7334,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.163265306122449e-05,
+      "loss": 0.6421,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.173469387755102e-05,
+      "loss": 0.7866,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1836734693877552e-05,
+      "loss": 0.7197,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1938775510204082e-05,
+      "loss": 0.6465,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2040816326530612e-05,
+      "loss": 0.7192,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2142857142857144e-05,
+      "loss": 0.5898,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2244897959183674e-05,
+      "loss": 0.6411,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2346938775510204e-05,
+      "loss": 0.6255,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2448979591836736e-05,
+      "loss": 0.6506,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2551020408163267e-05,
+      "loss": 0.6919,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2653061224489797e-05,
+      "loss": 0.6196,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2755102040816325e-05,
+      "loss": 0.6807,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2857142857142857e-05,
+      "loss": 0.6265,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2959183673469387e-05,
+      "loss": 0.7461,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3061224489795918e-05,
+      "loss": 0.6125,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.316326530612245e-05,
+      "loss": 0.6897,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.326530612244898e-05,
+      "loss": 0.5996,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.336734693877551e-05,
+      "loss": 0.72,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3469387755102042e-05,
+      "loss": 0.6689,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3571428571428572e-05,
+      "loss": 0.6375,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3673469387755102e-05,
+      "loss": 0.6462,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3775510204081633e-05,
+      "loss": 0.7129,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3877551020408165e-05,
+      "loss": 0.7065,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3979591836734695e-05,
+      "loss": 0.7275,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4081632653061225e-05,
+      "loss": 0.6211,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4183673469387757e-05,
+      "loss": 0.677,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 0.6919,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4387755102040816e-05,
+      "loss": 0.6958,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4489795918367348e-05,
+      "loss": 0.5869,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4591836734693878e-05,
+      "loss": 0.7,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4693877551020408e-05,
+      "loss": 0.5664,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4795918367346938e-05,
+      "loss": 0.626,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.489795918367347e-05,
+      "loss": 0.6741,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5e-05,
+      "loss": 0.5891,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.510204081632653e-05,
+      "loss": 0.6362,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5204081632653063e-05,
+      "loss": 0.6792,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.530612244897959e-05,
+      "loss": 0.6304,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5408163265306123e-05,
+      "loss": 0.5828,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5510204081632655e-05,
+      "loss": 0.5793,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5612244897959183e-05,
+      "loss": 0.6584,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5714285714285715e-05,
+      "loss": 0.6138,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5816326530612247e-05,
+      "loss": 0.5945,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5918367346938776e-05,
+      "loss": 0.686,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6020408163265308e-05,
+      "loss": 0.6506,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.612244897959184e-05,
+      "loss": 0.6111,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6224489795918368e-05,
+      "loss": 0.6055,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.63265306122449e-05,
+      "loss": 0.7556,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6428571428571432e-05,
+      "loss": 0.6689,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6530612244897957e-05,
+      "loss": 0.6641,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.663265306122449e-05,
+      "loss": 0.627,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.673469387755102e-05,
+      "loss": 0.573,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.683673469387755e-05,
+      "loss": 0.5874,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.693877551020408e-05,
+      "loss": 0.7266,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.704081632653061e-05,
+      "loss": 0.5133,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 0.606,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7244897959183674e-05,
+      "loss": 0.7466,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7346938775510202e-05,
+      "loss": 0.6714,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7448979591836734e-05,
+      "loss": 0.5767,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7551020408163266e-05,
+      "loss": 0.6082,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7653061224489795e-05,
+      "loss": 0.6313,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7755102040816327e-05,
+      "loss": 0.6826,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.785714285714286e-05,
+      "loss": 0.6575,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7959183673469387e-05,
+      "loss": 0.563,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.806122448979592e-05,
+      "loss": 0.6057,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.816326530612245e-05,
+      "loss": 0.6367,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.826530612244898e-05,
+      "loss": 0.6848,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.836734693877551e-05,
+      "loss": 0.6113,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8469387755102043e-05,
+      "loss": 0.5989,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8571428571428572e-05,
+      "loss": 0.6594,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8673469387755104e-05,
+      "loss": 0.6526,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8775510204081636e-05,
+      "loss": 0.7329,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8877551020408164e-05,
+      "loss": 0.6379,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8979591836734696e-05,
+      "loss": 0.5686,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9081632653061225e-05,
+      "loss": 0.6121,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9183673469387756e-05,
+      "loss": 0.5588,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.928571428571429e-05,
+      "loss": 0.7256,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9387755102040817e-05,
+      "loss": 0.6768,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.948979591836735e-05,
+      "loss": 0.54,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9591836734693877e-05,
+      "loss": 0.5857,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9693877551020406e-05,
+      "loss": 0.6484,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9795918367346938e-05,
+      "loss": 0.781,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.989795918367347e-05,
+      "loss": 0.6611,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9999999999999998e-05,
+      "loss": 0.6052,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.010204081632653e-05,
+      "loss": 0.686,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0204081632653062e-05,
+      "loss": 0.6543,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.030612244897959e-05,
+      "loss": 0.6169,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0408163265306123e-05,
+      "loss": 0.7295,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0510204081632654e-05,
+      "loss": 0.6362,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0612244897959183e-05,
+      "loss": 0.625,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0714285714285715e-05,
+      "loss": 0.5625,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0816326530612247e-05,
+      "loss": 0.5674,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0918367346938775e-05,
+      "loss": 0.5908,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1020408163265307e-05,
+      "loss": 0.6963,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1122448979591836e-05,
+      "loss": 0.6631,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1224489795918368e-05,
+      "loss": 0.5596,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.13265306122449e-05,
+      "loss": 0.6018,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1428571428571428e-05,
+      "loss": 0.5964,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.153061224489796e-05,
+      "loss": 0.696,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1632653061224492e-05,
+      "loss": 0.6189,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.173469387755102e-05,
+      "loss": 0.6067,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1836734693877552e-05,
+      "loss": 0.6848,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1938775510204084e-05,
+      "loss": 0.689,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2040816326530613e-05,
+      "loss": 0.6631,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2142857142857145e-05,
+      "loss": 0.6411,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2244897959183677e-05,
+      "loss": 0.6821,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2346938775510205e-05,
+      "loss": 0.647,
+      "step": 219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2448979591836737e-05,
+      "loss": 0.5479,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2551020408163266e-05,
+      "loss": 0.5742,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2653061224489794e-05,
+      "loss": 0.7515,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2755102040816326e-05,
+      "loss": 0.6558,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2857142857142858e-05,
+      "loss": 0.6619,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2959183673469387e-05,
+      "loss": 0.6653,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.306122448979592e-05,
+      "loss": 0.6743,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3163265306122447e-05,
+      "loss": 0.5967,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.326530612244898e-05,
+      "loss": 0.6167,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.336734693877551e-05,
+      "loss": 0.6992,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.346938775510204e-05,
+      "loss": 0.5959,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.357142857142857e-05,
+      "loss": 0.6304,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3673469387755103e-05,
+      "loss": 0.6453,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3775510204081632e-05,
+      "loss": 0.6145,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3877551020408164e-05,
+      "loss": 0.6633,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3979591836734696e-05,
+      "loss": 0.6887,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4081632653061224e-05,
+      "loss": 0.6138,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4183673469387756e-05,
+      "loss": 0.6514,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4285714285714288e-05,
+      "loss": 0.5928,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4387755102040816e-05,
+      "loss": 0.6228,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.448979591836735e-05,
+      "loss": 0.5391,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.459183673469388e-05,
+      "loss": 0.6719,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.469387755102041e-05,
+      "loss": 0.5977,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.479591836734694e-05,
+      "loss": 0.5015,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4897959183673473e-05,
+      "loss": 0.5813,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5212,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5102040816326533e-05,
+      "loss": 0.5671,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.520408163265306e-05,
+      "loss": 0.5916,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5306122448979594e-05,
+      "loss": 0.677,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5408163265306125e-05,
+      "loss": 0.5657,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.551020408163265e-05,
+      "loss": 0.6445,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5612244897959183e-05,
+      "loss": 0.6885,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5714285714285714e-05,
+      "loss": 0.5889,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5816326530612243e-05,
+      "loss": 0.6938,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5918367346938775e-05,
+      "loss": 0.6116,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6020408163265307e-05,
+      "loss": 0.5923,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6122448979591835e-05,
+      "loss": 0.6714,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6224489795918367e-05,
+      "loss": 0.5881,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.63265306122449e-05,
+      "loss": 0.6143,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6428571428571428e-05,
+      "loss": 0.6243,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.653061224489796e-05,
+      "loss": 0.5957,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.663265306122449e-05,
+      "loss": 0.583,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.673469387755102e-05,
+      "loss": 0.6807,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6836734693877552e-05,
+      "loss": 0.5723,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6938775510204084e-05,
+      "loss": 0.6833,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.7040816326530612e-05,
+      "loss": 0.6545,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.7142857142857144e-05,
+      "loss": 0.5549,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.7244897959183673e-05,
+      "loss": 0.6377,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.7346938775510205e-05,
+      "loss": 0.5762,
+      "step": 268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7448979591836737e-05,
+      "loss": 0.6821,
+      "step": 269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7551020408163265e-05,
+      "loss": 0.615,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7653061224489797e-05,
+      "loss": 0.6445,
+      "step": 271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.775510204081633e-05,
+      "loss": 0.5815,
+      "step": 272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7857142857142858e-05,
+      "loss": 0.5498,
+      "step": 273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.795918367346939e-05,
+      "loss": 0.7029,
+      "step": 274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.806122448979592e-05,
+      "loss": 0.6799,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.816326530612245e-05,
+      "loss": 0.5461,
+      "step": 276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8265306122448982e-05,
+      "loss": 0.6794,
+      "step": 277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8367346938775514e-05,
+      "loss": 0.574,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8469387755102042e-05,
+      "loss": 0.7373,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 0.6125,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8673469387755103e-05,
+      "loss": 0.6726,
+      "step": 281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.877551020408163e-05,
+      "loss": 0.5234,
+      "step": 282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8877551020408163e-05,
+      "loss": 0.5254,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8979591836734695e-05,
+      "loss": 0.5603,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9081632653061224e-05,
+      "loss": 0.553,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9183673469387756e-05,
+      "loss": 0.6526,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9285714285714284e-05,
+      "loss": 0.6494,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9387755102040816e-05,
+      "loss": 0.5557,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9489795918367348e-05,
+      "loss": 0.6804,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9591836734693876e-05,
+      "loss": 0.613,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.969387755102041e-05,
+      "loss": 0.7083,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.979591836734694e-05,
+      "loss": 0.6125,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.989795918367347e-05,
+      "loss": 0.6155,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3e-05,
+      "loss": 0.5828,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999999175303322e-05,
+      "loss": 0.561,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999999670121339e-05,
+      "loss": 0.6365,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999992577730462e-05,
+      "loss": 0.5732,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999986804855002e-05,
+      "loss": 0.6289,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999997938258764e-05,
+      "loss": 0.5459,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999970310929192e-05,
+      "loss": 0.5874,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999995958988066e-05,
+      "loss": 0.5466,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999947219443217e-05,
+      "loss": 0.6543,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999993319961823e-05,
+      "loss": 0.6423,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999991753040723e-05,
+      "loss": 0.6096,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999900211811956e-05,
+      "loss": 0.5486,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999881243834292e-05,
+      "loss": 0.5874,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999860626476347e-05,
+      "loss": 0.5869,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999838359740362e-05,
+      "loss": 0.6101,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999814443628812e-05,
+      "loss": 0.6245,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999978887814431e-05,
+      "loss": 0.627,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999761663289668e-05,
+      "loss": 0.5627,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999973279906788e-05,
+      "loss": 0.5471,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999970228548213e-05,
+      "loss": 0.6611,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999670122535755e-05,
+      "loss": 0.5405,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999636310232314e-05,
+      "loss": 0.5759,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999600848575503e-05,
+      "loss": 0.5969,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999563737569235e-05,
+      "loss": 0.6221,
+      "step": 317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9999524977217583e-05,
+      "loss": 0.6707,
+      "step": 318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9999484567524813e-05,
+      "loss": 0.5391,
+      "step": 319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9999442508495367e-05,
+      "loss": 0.6521,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9999398800133875e-05,
+      "loss": 0.6309,
+      "step": 321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999935344244514e-05,
+      "loss": 0.574,
+      "step": 322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999930643543414e-05,
+      "loss": 0.603,
+      "step": 323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9999257779106063e-05,
+      "loss": 0.6025,
+      "step": 324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9999207473466242e-05,
+      "loss": 0.5771,
+      "step": 325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999915551852022e-05,
+      "loss": 0.6641,
+      "step": 326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.99991019142737e-05,
+      "loss": 0.5759,
+      "step": 327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9999046660732585e-05,
+      "loss": 0.6323,
+      "step": 328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998989757902945e-05,
+      "loss": 0.6282,
+      "step": 329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998931205791042e-05,
+      "loss": 0.595,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999887100440331e-05,
+      "loss": 0.5364,
+      "step": 331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998809153746366e-05,
+      "loss": 0.5558,
+      "step": 332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999874565382702e-05,
+      "loss": 0.6003,
+      "step": 333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999868050465225e-05,
+      "loss": 0.5276,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998613706229217e-05,
+      "loss": 0.6484,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998545258565275e-05,
+      "loss": 0.6028,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998475161667936e-05,
+      "loss": 0.5464,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998403415544917e-05,
+      "loss": 0.5542,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998330020204113e-05,
+      "loss": 0.5972,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998254975653577e-05,
+      "loss": 0.4961,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999817828190158e-05,
+      "loss": 0.7126,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998099938956545e-05,
+      "loss": 0.6094,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998019946827084e-05,
+      "loss": 0.573,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997938305522003e-05,
+      "loss": 0.6008,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997855015050268e-05,
+      "loss": 0.6238,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997770075421044e-05,
+      "loss": 0.502,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997683486643673e-05,
+      "loss": 0.6038,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999759524872767e-05,
+      "loss": 0.5703,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997505361682742e-05,
+      "loss": 0.658,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997413825518772e-05,
+      "loss": 0.5256,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997320640245828e-05,
+      "loss": 0.6078,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997225805874146e-05,
+      "loss": 0.5439,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997129322414168e-05,
+      "loss": 0.687,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997031189876495e-05,
+      "loss": 0.6265,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996931408271916e-05,
+      "loss": 0.6348,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996829977611414e-05,
+      "loss": 0.7388,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996726897906127e-05,
+      "loss": 0.5842,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.99966221691674e-05,
+      "loss": 0.5605,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996515791406746e-05,
+      "loss": 0.5532,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996407764635865e-05,
+      "loss": 0.6357,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996298088866632e-05,
+      "loss": 0.5374,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996186764111103e-05,
+      "loss": 0.5515,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996073790381532e-05,
+      "loss": 0.5886,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9995959167690328e-05,
+      "loss": 0.584,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9995842896050103e-05,
+      "loss": 0.4829,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9995724975473636e-05,
+      "loss": 0.6343,
+      "step": 366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.99956054059739e-05,
+      "loss": 0.5906,
+      "step": 367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9995484187564042e-05,
+      "loss": 0.5479,
+      "step": 368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9995361320257385e-05,
+      "loss": 0.5686,
+      "step": 369
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9995236804067447e-05,
+      "loss": 0.6067,
+      "step": 370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9995110639007914e-05,
+      "loss": 0.4718,
+      "step": 371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999498282509266e-05,
+      "loss": 0.5251,
+      "step": 372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9994853362335742e-05,
+      "loss": 0.5435,
+      "step": 373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9994722250751395e-05,
+      "loss": 0.6294,
+      "step": 374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999458949035404e-05,
+      "loss": 0.6147,
+      "step": 375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999445508115826e-05,
+      "loss": 0.6516,
+      "step": 376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9994319023178857e-05,
+      "loss": 0.5627,
+      "step": 377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999418131643077e-05,
+      "loss": 0.532,
+      "step": 378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9994041960929155e-05,
+      "loss": 0.5532,
+      "step": 379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9993900956689335e-05,
+      "loss": 0.5774,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9993758303726807e-05,
+      "loss": 0.5461,
+      "step": 381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9993614002057262e-05,
+      "loss": 0.5962,
+      "step": 382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9993468051696568e-05,
+      "loss": 0.6218,
+      "step": 383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999332045266077e-05,
+      "loss": 0.5708,
+      "step": 384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9993171204966107e-05,
+      "loss": 0.5625,
+      "step": 385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999302030862898e-05,
+      "loss": 0.5618,
+      "step": 386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999286776366598e-05,
+      "loss": 0.6147,
+      "step": 387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9992713570093896e-05,
+      "loss": 0.594,
+      "step": 388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9992557727929663e-05,
+      "loss": 0.6042,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9992400237190433e-05,
+      "loss": 0.519,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9992241097893518e-05,
+      "loss": 0.4951,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9992080310056413e-05,
+      "loss": 0.5886,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9991917873696802e-05,
+      "loss": 0.6218,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9991753788832552e-05,
+      "loss": 0.4844,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9991588055481698e-05,
+      "loss": 0.634,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9991420673662464e-05,
+      "loss": 0.6099,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9991251643393258e-05,
+      "loss": 0.6467,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9991080964692664e-05,
+      "loss": 0.5879,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9990908637579453e-05,
+      "loss": 0.5081,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9990734662072577e-05,
+      "loss": 0.5627,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9990559038191155e-05,
+      "loss": 0.6348,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999038176595451e-05,
+      "loss": 0.5396,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9990202845382126e-05,
+      "loss": 0.5024,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9990022276493685e-05,
+      "loss": 0.5261,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998984005930904e-05,
+      "loss": 0.553,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998965619384822e-05,
+      "loss": 0.4834,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9989470680131454e-05,
+      "loss": 0.5493,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998928351817913e-05,
+      "loss": 0.6875,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9989094708011836e-05,
+      "loss": 0.6357,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998890424965033e-05,
+      "loss": 0.5635,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998871214311556e-05,
+      "loss": 0.5496,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9988518388428644e-05,
+      "loss": 0.5991,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998832298561089e-05,
+      "loss": 0.5007,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9988125934683788e-05,
+      "loss": 0.4937,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9987927235668996e-05,
+      "loss": 0.5022,
+      "step": 415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998772688858837e-05,
+      "loss": 0.6116,
+      "step": 416
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9987524893463935e-05,
+      "loss": 0.6221,
+      "step": 417
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9987321250317908e-05,
+      "loss": 0.5603,
+      "step": 418
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998711595917268e-05,
+      "loss": 0.5605,
+      "step": 419
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9986909020050822e-05,
+      "loss": 0.5544,
+      "step": 420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9986700432975093e-05,
+      "loss": 0.5212,
+      "step": 421
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9986490197968426e-05,
+      "loss": 0.5129,
+      "step": 422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998627831505394e-05,
+      "loss": 0.6321,
+      "step": 423
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998606478425493e-05,
+      "loss": 0.6152,
+      "step": 424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998584960559488e-05,
+      "loss": 0.5569,
+      "step": 425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998563277909745e-05,
+      "loss": 0.553,
+      "step": 426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998541430478648e-05,
+      "loss": 0.7063,
+      "step": 427
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998519418268599e-05,
+      "loss": 0.5769,
+      "step": 428
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.99849724128202e-05,
+      "loss": 0.561,
+      "step": 429
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998474899521348e-05,
+      "loss": 0.5837,
+      "step": 430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.99845239298904e-05,
+      "loss": 0.574,
+      "step": 431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9984297216875714e-05,
+      "loss": 0.5237,
+      "step": 432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998406885619435e-05,
+      "loss": 0.6077,
+      "step": 433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998383884787141e-05,
+      "loss": 0.5967,
+      "step": 434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9983607191932192e-05,
+      "loss": 0.6074,
+      "step": 435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998337388840217e-05,
+      "loss": 0.541,
+      "step": 436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9983138937306996e-05,
+      "loss": 0.5242,
+      "step": 437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9982902338672506e-05,
+      "loss": 0.4312,
+      "step": 438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9982664092524714e-05,
+      "loss": 0.5576,
+      "step": 439
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9982424198889823e-05,
+      "loss": 0.5581,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9982182657794206e-05,
+      "loss": 0.4971,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998193946926442e-05,
+      "loss": 0.4963,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9981694633327214e-05,
+      "loss": 0.5488,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9981448150009506e-05,
+      "loss": 0.6221,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9981200019338403e-05,
+      "loss": 0.5554,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998095024134118e-05,
+      "loss": 0.5806,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998069881604531e-05,
+      "loss": 0.4768,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9980445743478438e-05,
+      "loss": 0.5144,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9980191023668394e-05,
+      "loss": 0.5862,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9979934656643184e-05,
+      "loss": 0.4839,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9979676642430995e-05,
+      "loss": 0.5923,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9979416981060204e-05,
+      "loss": 0.5728,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9979155672559362e-05,
+      "loss": 0.4453,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.99788927169572e-05,
+      "loss": 0.5942,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9978628114282635e-05,
+      "loss": 0.5039,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.997836186456476e-05,
+      "loss": 0.498,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9978093967832857e-05,
+      "loss": 0.4636,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9977824424116378e-05,
+      "loss": 0.5239,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9977553233444957e-05,
+      "loss": 0.4563,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.997728039584843e-05,
+      "loss": 0.5649,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9977005911356787e-05,
+      "loss": 0.5374,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.997672978000021e-05,
+      "loss": 0.4443,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9976452001809065e-05,
+      "loss": 0.4988,
+      "step": 463
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9976172576813895e-05,
+      "loss": 0.5493,
+      "step": 464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997589150504543e-05,
+      "loss": 0.5212,
+      "step": 465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997560878653457e-05,
+      "loss": 0.4868,
+      "step": 466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9975324421312406e-05,
+      "loss": 0.6072,
+      "step": 467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9975038409410205e-05,
+      "loss": 0.5552,
+      "step": 468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9974750750859417e-05,
+      "loss": 0.4897,
+      "step": 469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9974461445691673e-05,
+      "loss": 0.5601,
+      "step": 470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997417049393879e-05,
+      "loss": 0.6023,
+      "step": 471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997387789563275e-05,
+      "loss": 0.5144,
+      "step": 472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997358365080574e-05,
+      "loss": 0.5854,
+      "step": 473
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9973287759490108e-05,
+      "loss": 0.5469,
+      "step": 474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997299022171839e-05,
+      "loss": 0.5757,
+      "step": 475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9972691037523303e-05,
+      "loss": 0.5903,
+      "step": 476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9972390206937742e-05,
+      "loss": 0.4905,
+      "step": 477
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997208772999479e-05,
+      "loss": 0.4932,
+      "step": 478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997178360672771e-05,
+      "loss": 0.405,
+      "step": 479
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9971477837169943e-05,
+      "loss": 0.491,
+      "step": 480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9971170421355106e-05,
+      "loss": 0.6345,
+      "step": 481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9970861359317002e-05,
+      "loss": 0.5054,
+      "step": 482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9970550651089623e-05,
+      "loss": 0.5288,
+      "step": 483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997023829670713e-05,
+      "loss": 0.6409,
+      "step": 484
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9969924296203866e-05,
+      "loss": 0.5183,
+      "step": 485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9969608649614362e-05,
+      "loss": 0.5664,
+      "step": 486
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9969291356973325e-05,
+      "loss": 0.4514,
+      "step": 487
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9968972418315648e-05,
+      "loss": 0.605,
+      "step": 488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.99686518336764e-05,
+      "loss": 0.5442,
+      "step": 489
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9968329603090825e-05,
+      "loss": 0.4917,
+      "step": 490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9968005726594365e-05,
+      "loss": 0.4777,
+      "step": 491
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9967680204222628e-05,
+      "loss": 0.5669,
+      "step": 492
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9967353036011408e-05,
+      "loss": 0.4692,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9967024221996685e-05,
+      "loss": 0.5276,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9966693762214615e-05,
+      "loss": 0.5073,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.996636165670153e-05,
+      "loss": 0.5674,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9966027905493946e-05,
+      "loss": 0.541,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9965692508628572e-05,
+      "loss": 0.6008,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9965355466142278e-05,
+      "loss": 0.5598,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9965016778072135e-05,
+      "loss": 0.5483,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.996467644445538e-05,
+      "loss": 0.6516,
+      "step": 501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9964334465329435e-05,
+      "loss": 0.5491,
+      "step": 502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9963990840731902e-05,
+      "loss": 0.5173,
+      "step": 503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9963645570700573e-05,
+      "loss": 0.5701,
+      "step": 504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9963298655273405e-05,
+      "loss": 0.5288,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9962950094488552e-05,
+      "loss": 0.5757,
+      "step": 506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9962599888384334e-05,
+      "loss": 0.4807,
+      "step": 507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.996224803699927e-05,
+      "loss": 0.4948,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.996189454037204e-05,
+      "loss": 0.553,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.996153939854152e-05,
+      "loss": 0.5063,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9961182611546757e-05,
+      "loss": 0.5107,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9960824179426984e-05,
+      "loss": 0.5247,
+      "step": 512
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9960464102221617e-05,
+      "loss": 0.5444,
+      "step": 513
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.996010237997025e-05,
+      "loss": 0.5176,
+      "step": 514
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9959739012712657e-05,
+      "loss": 0.5176,
+      "step": 515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9959374000488788e-05,
+      "loss": 0.4741,
+      "step": 516
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9959007343338786e-05,
+      "loss": 0.5686,
+      "step": 517
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9958639041302968e-05,
+      "loss": 0.5552,
+      "step": 518
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.995826909442183e-05,
+      "loss": 0.5059,
+      "step": 519
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.995789750273605e-05,
+      "loss": 0.5127,
+      "step": 520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9957524266286496e-05,
+      "loss": 0.5789,
+      "step": 521
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9957149385114198e-05,
+      "loss": 0.4744,
+      "step": 522
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.995677285926038e-05,
+      "loss": 0.4097,
+      "step": 523
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9956394688766454e-05,
+      "loss": 0.4844,
+      "step": 524
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9956014873673998e-05,
+      "loss": 0.5264,
+      "step": 525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.995563341402477e-05,
+      "loss": 0.4753,
+      "step": 526
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9955250309860723e-05,
+      "loss": 0.5847,
+      "step": 527
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.995486556122398e-05,
+      "loss": 0.5366,
+      "step": 528
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9954479168156852e-05,
+      "loss": 0.4768,
+      "step": 529
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9954091130701816e-05,
+      "loss": 0.4927,
+      "step": 530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9953701448901557e-05,
+      "loss": 0.5127,
+      "step": 531
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9953310122798907e-05,
+      "loss": 0.5334,
+      "step": 532
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9952917152436903e-05,
+      "loss": 0.4868,
+      "step": 533
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.995252253785876e-05,
+      "loss": 0.4622,
+      "step": 534
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9952126279107865e-05,
+      "loss": 0.4221,
+      "step": 535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9951728376227793e-05,
+      "loss": 0.5464,
+      "step": 536
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9951328829262293e-05,
+      "loss": 0.448,
+      "step": 537
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.99509276382553e-05,
+      "loss": 0.5073,
+      "step": 538
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9950524803250935e-05,
+      "loss": 0.5032,
+      "step": 539
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9950120324293484e-05,
+      "loss": 0.4229,
+      "step": 540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994971420142743e-05,
+      "loss": 0.4844,
+      "step": 541
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994930643469743e-05,
+      "loss": 0.4453,
+      "step": 542
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9948897024148326e-05,
+      "loss": 0.5215,
+      "step": 543
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994848596982512e-05,
+      "loss": 0.5115,
+      "step": 544
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994807327177303e-05,
+      "loss": 0.6145,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9947658930037426e-05,
+      "loss": 0.4913,
+      "step": 546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9947242944663874e-05,
+      "loss": 0.5344,
+      "step": 547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9946825315698105e-05,
+      "loss": 0.5339,
+      "step": 548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9946406043186054e-05,
+      "loss": 0.6082,
+      "step": 549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994598512717382e-05,
+      "loss": 0.4326,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9945562567707684e-05,
+      "loss": 0.5732,
+      "step": 551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994513836483412e-05,
+      "loss": 0.5239,
+      "step": 552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9944712518599755e-05,
+      "loss": 0.5632,
+      "step": 553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9944285029051427e-05,
+      "loss": 0.4907,
+      "step": 554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9943855896236146e-05,
+      "loss": 0.4775,
+      "step": 555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9943425120201087e-05,
+      "loss": 0.4648,
+      "step": 556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9942992700993628e-05,
+      "loss": 0.5977,
+      "step": 557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9942558638661316e-05,
+      "loss": 0.5125,
+      "step": 558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9942122933251878e-05,
+      "loss": 0.5132,
+      "step": 559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9941685584813228e-05,
+      "loss": 0.5791,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994124659339345e-05,
+      "loss": 0.4939,
+      "step": 561
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.994080595904082e-05,
+      "loss": 0.4717,
+      "step": 562
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.994036368180379e-05,
+      "loss": 0.5522,
+      "step": 563
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9939919761730993e-05,
+      "loss": 0.4846,
+      "step": 564
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9939474198871236e-05,
+      "loss": 0.4252,
+      "step": 565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9939026993273518e-05,
+      "loss": 0.4299,
+      "step": 566
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9938578144987017e-05,
+      "loss": 0.604,
+      "step": 567
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9938127654061082e-05,
+      "loss": 0.4963,
+      "step": 568
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9937675520545248e-05,
+      "loss": 0.5667,
+      "step": 569
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.993722174448924e-05,
+      "loss": 0.5476,
+      "step": 570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9936766325942952e-05,
+      "loss": 0.5457,
+      "step": 571
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.993630926495645e-05,
+      "loss": 0.5476,
+      "step": 572
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.993585056158001e-05,
+      "loss": 0.5566,
+      "step": 573
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9935390215864057e-05,
+      "loss": 0.4343,
+      "step": 574
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9934928227859215e-05,
+      "loss": 0.5386,
+      "step": 575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9934464597616287e-05,
+      "loss": 0.4956,
+      "step": 576
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.993399932518625e-05,
+      "loss": 0.5356,
+      "step": 577
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9933532410620266e-05,
+      "loss": 0.4624,
+      "step": 578
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.993306385396968e-05,
+      "loss": 0.452,
+      "step": 579
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9932593655286006e-05,
+      "loss": 0.4762,
+      "step": 580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.993212181462096e-05,
+      "loss": 0.5701,
+      "step": 581
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9931648332026407e-05,
+      "loss": 0.5063,
+      "step": 582
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9931173207554428e-05,
+      "loss": 0.5327,
+      "step": 583
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.993069644125726e-05,
+      "loss": 0.4335,
+      "step": 584
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.993021803318733e-05,
+      "loss": 0.4575,
+      "step": 585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.992973798339724e-05,
+      "loss": 0.5752,
+      "step": 586
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9929256291939783e-05,
+      "loss": 0.4441,
+      "step": 587
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9928772958867917e-05,
+      "loss": 0.4768,
+      "step": 588
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9928287984234796e-05,
+      "loss": 0.3717,
+      "step": 589
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9927801368093744e-05,
+      "loss": 0.5181,
+      "step": 590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9927313110498266e-05,
+      "loss": 0.4856,
+      "step": 591
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9926823211502063e-05,
+      "loss": 0.4585,
+      "step": 592
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9926331671158987e-05,
+      "loss": 0.5178,
+      "step": 593
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.99258384895231e-05,
+      "loss": 0.5681,
+      "step": 594
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.992534366664863e-05,
+      "loss": 0.4995,
+      "step": 595
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.992484720258998e-05,
+      "loss": 0.4148,
+      "step": 596
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.992434909740175e-05,
+      "loss": 0.498,
+      "step": 597
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.992384935113871e-05,
+      "loss": 0.4827,
+      "step": 598
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9923347963855806e-05,
+      "loss": 0.4271,
+      "step": 599
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9922844935608177e-05,
+      "loss": 0.4253,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9922340266451133e-05,
+      "loss": 0.5416,
+      "step": 601
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9921833956440167e-05,
+      "loss": 0.4608,
+      "step": 602
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9921326005630952e-05,
+      "loss": 0.4893,
+      "step": 603
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9920816414079344e-05,
+      "loss": 0.4673,
+      "step": 604
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9920305181841374e-05,
+      "loss": 0.5381,
+      "step": 605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9919792308973264e-05,
+      "loss": 0.4917,
+      "step": 606
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.99192777955314e-05,
+      "loss": 0.4788,
+      "step": 607
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9918761641572365e-05,
+      "loss": 0.5081,
+      "step": 608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.991824384715291e-05,
+      "loss": 0.4419,
+      "step": 609
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9917724412329973e-05,
+      "loss": 0.4707,
+      "step": 610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9917203337160672e-05,
+      "loss": 0.4423,
+      "step": 611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9916680621702308e-05,
+      "loss": 0.4299,
+      "step": 612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.991615626601235e-05,
+      "loss": 0.5149,
+      "step": 613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.991563027014846e-05,
+      "loss": 0.4536,
+      "step": 614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9915102634168482e-05,
+      "loss": 0.4692,
+      "step": 615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9914573358130425e-05,
+      "loss": 0.4282,
+      "step": 616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.991404244209249e-05,
+      "loss": 0.5,
+      "step": 617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.991350988611306e-05,
+      "loss": 0.4359,
+      "step": 618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9912975690250694e-05,
+      "loss": 0.4482,
+      "step": 619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.991243985456413e-05,
+      "loss": 0.5181,
+      "step": 620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.991190237911229e-05,
+      "loss": 0.5251,
+      "step": 621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9911363263954272e-05,
+      "loss": 0.4529,
+      "step": 622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9910822509149362e-05,
+      "loss": 0.4822,
+      "step": 623
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9910280114757012e-05,
+      "loss": 0.4673,
+      "step": 624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9909736080836875e-05,
+      "loss": 0.4841,
+      "step": 625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.990919040744876e-05,
+      "loss": 0.5261,
+      "step": 626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.990864309465268e-05,
+      "loss": 0.5093,
+      "step": 627
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9908094142508817e-05,
+      "loss": 0.4998,
+      "step": 628
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9907543551077527e-05,
+      "loss": 0.4341,
+      "step": 629
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.990699132041935e-05,
+      "loss": 0.5442,
+      "step": 630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9906437450595023e-05,
+      "loss": 0.4736,
+      "step": 631
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9905881941665436e-05,
+      "loss": 0.5157,
+      "step": 632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.990532479369168e-05,
+      "loss": 0.4617,
+      "step": 633
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9904766006735012e-05,
+      "loss": 0.5095,
+      "step": 634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9904205580856883e-05,
+      "loss": 0.5122,
+      "step": 635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9903643516118913e-05,
+      "loss": 0.3665,
+      "step": 636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.990307981258291e-05,
+      "loss": 0.4797,
+      "step": 637
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.990251447031085e-05,
+      "loss": 0.4473,
+      "step": 638
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.990194748936491e-05,
+      "loss": 0.5273,
+      "step": 639
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9901378869807428e-05,
+      "loss": 0.537,
+      "step": 640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.990080861170093e-05,
+      "loss": 0.5479,
+      "step": 641
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.990023671510812e-05,
+      "loss": 0.5344,
+      "step": 642
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9899663180091886e-05,
+      "loss": 0.49,
+      "step": 643
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9899088006715294e-05,
+      "loss": 0.6523,
+      "step": 644
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.989851119504159e-05,
+      "loss": 0.4851,
+      "step": 645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.989793274513419e-05,
+      "loss": 0.4917,
+      "step": 646
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9897352657056716e-05,
+      "loss": 0.4164,
+      "step": 647
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9896770930872945e-05,
+      "loss": 0.4915,
+      "step": 648
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9896187566646845e-05,
+      "loss": 0.4641,
+      "step": 649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.989560256444256e-05,
+      "loss": 0.5698,
+      "step": 650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9895015924324424e-05,
+      "loss": 0.5232,
+      "step": 651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9894427646356934e-05,
+      "loss": 0.542,
+      "step": 652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9893837730604785e-05,
+      "loss": 0.4429,
+      "step": 653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9893246177132837e-05,
+      "loss": 0.5144,
+      "step": 654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9892652986006146e-05,
+      "loss": 0.454,
+      "step": 655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.989205815728993e-05,
+      "loss": 0.5273,
+      "step": 656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9891461691049597e-05,
+      "loss": 0.4995,
+      "step": 657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.989086358735074e-05,
+      "loss": 0.5483,
+      "step": 658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.989026384625912e-05,
+      "loss": 0.4783,
+      "step": 659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.988966246784069e-05,
+      "loss": 0.4648,
+      "step": 660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9889059452161572e-05,
+      "loss": 0.4337,
+      "step": 661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.988845479928808e-05,
+      "loss": 0.4834,
+      "step": 662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9887848509286694e-05,
+      "loss": 0.392,
+      "step": 663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9887240582224085e-05,
+      "loss": 0.3989,
+      "step": 664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9886631018167103e-05,
+      "loss": 0.3881,
+      "step": 665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9886019817182768e-05,
+      "loss": 0.4285,
+      "step": 666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9885406979338294e-05,
+      "loss": 0.4265,
+      "step": 667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9884792504701065e-05,
+      "loss": 0.4685,
+      "step": 668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.988417639333865e-05,
+      "loss": 0.3955,
+      "step": 669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9883558645318798e-05,
+      "loss": 0.4851,
+      "step": 670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.988293926070943e-05,
+      "loss": 0.4398,
+      "step": 671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.988231823957866e-05,
+      "loss": 0.3478,
+      "step": 672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.988169558199477e-05,
+      "loss": 0.4465,
+      "step": 673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9881071288026234e-05,
+      "loss": 0.5452,
+      "step": 674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.988044535774169e-05,
+      "loss": 0.4686,
+      "step": 675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.987981779120997e-05,
+      "loss": 0.4585,
+      "step": 676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9879188588500088e-05,
+      "loss": 0.5217,
+      "step": 677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9878557749681216e-05,
+      "loss": 0.3647,
+      "step": 678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9877925274822732e-05,
+      "loss": 0.4437,
+      "step": 679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9877291163994176e-05,
+      "loss": 0.4976,
+      "step": 680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.987665541726528e-05,
+      "loss": 0.4907,
+      "step": 681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.987601803470595e-05,
+      "loss": 0.4775,
+      "step": 682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9875379016386267e-05,
+      "loss": 0.5393,
+      "step": 683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.98747383623765e-05,
+      "loss": 0.4823,
+      "step": 684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9874096072747098e-05,
+      "loss": 0.4253,
+      "step": 685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9873452147568684e-05,
+      "loss": 0.4707,
+      "step": 686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9872806586912063e-05,
+      "loss": 0.4556,
+      "step": 687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.987215939084822e-05,
+      "loss": 0.4685,
+      "step": 688
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.987151055944833e-05,
+      "loss": 0.5258,
+      "step": 689
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9870860092783724e-05,
+      "loss": 0.3882,
+      "step": 690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9870207990925932e-05,
+      "loss": 0.4448,
+      "step": 691
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.986955425394666e-05,
+      "loss": 0.4377,
+      "step": 692
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9868898881917795e-05,
+      "loss": 0.5381,
+      "step": 693
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9868241874911402e-05,
+      "loss": 0.4551,
+      "step": 694
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9867583232999718e-05,
+      "loss": 0.4036,
+      "step": 695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9866922956255173e-05,
+      "loss": 0.478,
+      "step": 696
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.986626104475037e-05,
+      "loss": 0.4355,
+      "step": 697
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9865597498558086e-05,
+      "loss": 0.4889,
+      "step": 698
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9864932317751298e-05,
+      "loss": 0.4302,
+      "step": 699
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9864265502403134e-05,
+      "loss": 0.5154,
+      "step": 700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9863597052586927e-05,
+      "loss": 0.5286,
+      "step": 701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9862926968376173e-05,
+      "loss": 0.4807,
+      "step": 702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9862255249844562e-05,
+      "loss": 0.3801,
+      "step": 703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9861581897065947e-05,
+      "loss": 0.5288,
+      "step": 704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9860906910114375e-05,
+      "loss": 0.4878,
+      "step": 705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.986023028906407e-05,
+      "loss": 0.5146,
+      "step": 706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9859552033989423e-05,
+      "loss": 0.4907,
+      "step": 707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9858872144965023e-05,
+      "loss": 0.5242,
+      "step": 708
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.985819062206563e-05,
+      "loss": 0.4758,
+      "step": 709
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.985750746536618e-05,
+      "loss": 0.5469,
+      "step": 710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9856822674941794e-05,
+      "loss": 0.4231,
+      "step": 711
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.985613625086777e-05,
+      "loss": 0.5144,
+      "step": 712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.985544819321959e-05,
+      "loss": 0.3649,
+      "step": 713
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9854758502072907e-05,
+      "loss": 0.4556,
+      "step": 714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9854067177503566e-05,
+      "loss": 0.4015,
+      "step": 715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9853374219587583e-05,
+      "loss": 0.4011,
+      "step": 716
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9852679628401156e-05,
+      "loss": 0.4916,
+      "step": 717
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9851983404020657e-05,
+      "loss": 0.3835,
+      "step": 718
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9851285546522645e-05,
+      "loss": 0.4934,
+      "step": 719
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.985058605598386e-05,
+      "loss": 0.5039,
+      "step": 720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.984988493248121e-05,
+      "loss": 0.5352,
+      "step": 721
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9849182176091797e-05,
+      "loss": 0.4142,
+      "step": 722
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9848477786892894e-05,
+      "loss": 0.4929,
+      "step": 723
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9847771764961956e-05,
+      "loss": 0.4756,
+      "step": 724
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9847064110376612e-05,
+      "loss": 0.4299,
+      "step": 725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9846354823214683e-05,
+      "loss": 0.4661,
+      "step": 726
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9845643903554153e-05,
+      "loss": 0.4871,
+      "step": 727
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9844931351473202e-05,
+      "loss": 0.5452,
+      "step": 728
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9844217167050182e-05,
+      "loss": 0.4353,
+      "step": 729
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9843501350363614e-05,
+      "loss": 0.4473,
+      "step": 730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9842783901492224e-05,
+      "loss": 0.5234,
+      "step": 731
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9842064820514895e-05,
+      "loss": 0.3857,
+      "step": 732
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9841344107510692e-05,
+      "loss": 0.5073,
+      "step": 733
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.984062176255887e-05,
+      "loss": 0.4714,
+      "step": 734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9839897785738855e-05,
+      "loss": 0.4436,
+      "step": 735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9839172177130263e-05,
+      "loss": 0.4321,
+      "step": 736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9838444936812872e-05,
+      "loss": 0.5006,
+      "step": 737
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9837716064866655e-05,
+      "loss": 0.4081,
+      "step": 738
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.983698556137175e-05,
+      "loss": 0.4512,
+      "step": 739
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.983625342640849e-05,
+      "loss": 0.4946,
+      "step": 740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9835519660057385e-05,
+      "loss": 0.4739,
+      "step": 741
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.983478426239911e-05,
+      "loss": 0.4849,
+      "step": 742
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9834047233514533e-05,
+      "loss": 0.5028,
+      "step": 743
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9833308573484697e-05,
+      "loss": 0.4875,
+      "step": 744
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9832568282390826e-05,
+      "loss": 0.4282,
+      "step": 745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9831826360314316e-05,
+      "loss": 0.4788,
+      "step": 746
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9831082807336752e-05,
+      "loss": 0.5239,
+      "step": 747
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9830337623539905e-05,
+      "loss": 0.4004,
+      "step": 748
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.98295908090057e-05,
+      "loss": 0.4659,
+      "step": 749
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9828842363816266e-05,
+      "loss": 0.5864,
+      "step": 750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9828092288053894e-05,
+      "loss": 0.5576,
+      "step": 751
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9827340581801067e-05,
+      "loss": 0.454,
+      "step": 752
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9826587245140447e-05,
+      "loss": 0.4548,
+      "step": 753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.982583227815486e-05,
+      "loss": 0.4774,
+      "step": 754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9825075680927325e-05,
+      "loss": 0.4111,
+      "step": 755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9824317453541042e-05,
+      "loss": 0.5305,
+      "step": 756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.982355759607938e-05,
+      "loss": 0.3794,
+      "step": 757
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9822796108625898e-05,
+      "loss": 0.3912,
+      "step": 758
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9822032991264326e-05,
+      "loss": 0.4348,
+      "step": 759
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9821268244078572e-05,
+      "loss": 0.4268,
+      "step": 760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9820501867152737e-05,
+      "loss": 0.4415,
+      "step": 761
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.981973386057108e-05,
+      "loss": 0.4553,
+      "step": 762
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9818964224418057e-05,
+      "loss": 0.4026,
+      "step": 763
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9818192958778297e-05,
+      "loss": 0.442,
+      "step": 764
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.981742006373661e-05,
+      "loss": 0.3972,
+      "step": 765
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9816645539377975e-05,
+      "loss": 0.4377,
+      "step": 766
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9815869385787565e-05,
+      "loss": 0.5033,
+      "step": 767
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9815091603050725e-05,
+      "loss": 0.4248,
+      "step": 768
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9814312191252978e-05,
+      "loss": 0.3317,
+      "step": 769
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9813531150480028e-05,
+      "loss": 0.4199,
+      "step": 770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.981274848081776e-05,
+      "loss": 0.3796,
+      "step": 771
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9811964182352234e-05,
+      "loss": 0.4924,
+      "step": 772
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9811178255169692e-05,
+      "loss": 0.361,
+      "step": 773
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9810390699356552e-05,
+      "loss": 0.4468,
+      "step": 774
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.980960151499942e-05,
+      "loss": 0.5271,
+      "step": 775
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.980881070218506e-05,
+      "loss": 0.5044,
+      "step": 776
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9808018261000445e-05,
+      "loss": 0.4568,
+      "step": 777
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9807224191532707e-05,
+      "loss": 0.4954,
+      "step": 778
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9806428493869156e-05,
+      "loss": 0.4697,
+      "step": 779
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9805631168097293e-05,
+      "loss": 0.4207,
+      "step": 780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9804832214304788e-05,
+      "loss": 0.5083,
+      "step": 781
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.980403163257949e-05,
+      "loss": 0.4264,
+      "step": 782
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9803229423009442e-05,
+      "loss": 0.385,
+      "step": 783
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9802425585682845e-05,
+      "loss": 0.52,
+      "step": 784
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9801620120688094e-05,
+      "loss": 0.4951,
+      "step": 785
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9800813028113752e-05,
+      "loss": 0.4429,
+      "step": 786
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.980000430804857e-05,
+      "loss": 0.4143,
+      "step": 787
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9799193960581473e-05,
+      "loss": 0.4053,
+      "step": 788
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.979838198580157e-05,
+      "loss": 0.3644,
+      "step": 789
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.979756838379814e-05,
+      "loss": 0.4503,
+      "step": 790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9796753154660654e-05,
+      "loss": 0.4407,
+      "step": 791
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9795936298478746e-05,
+      "loss": 0.4635,
+      "step": 792
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9795117815342238e-05,
+      "loss": 0.5238,
+      "step": 793
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9794297705341136e-05,
+      "loss": 0.4976,
+      "step": 794
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9793475968565614e-05,
+      "loss": 0.449,
+      "step": 795
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9792652605106032e-05,
+      "loss": 0.5505,
+      "step": 796
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9791827615052926e-05,
+      "loss": 0.4679,
+      "step": 797
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.979100099849701e-05,
+      "loss": 0.4333,
+      "step": 798
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9790172755529182e-05,
+      "loss": 0.5232,
+      "step": 799
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.978934288624051e-05,
+      "loss": 0.4861,
+      "step": 800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9788511390722253e-05,
+      "loss": 0.4379,
+      "step": 801
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9787678269065836e-05,
+      "loss": 0.4268,
+      "step": 802
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9786843521362875e-05,
+      "loss": 0.4451,
+      "step": 803
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9786007147705153e-05,
+      "loss": 0.4377,
+      "step": 804
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9785169148184632e-05,
+      "loss": 0.4246,
+      "step": 805
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.978432952289347e-05,
+      "loss": 0.5295,
+      "step": 806
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9783488271923984e-05,
+      "loss": 0.4136,
+      "step": 807
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9782645395368686e-05,
+      "loss": 0.3757,
+      "step": 808
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.978180089332025e-05,
+      "loss": 0.4622,
+      "step": 809
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9780954765871537e-05,
+      "loss": 0.4893,
+      "step": 810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9780107013115586e-05,
+      "loss": 0.4088,
+      "step": 811
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9779257635145623e-05,
+      "loss": 0.552,
+      "step": 812
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.977840663205504e-05,
+      "loss": 0.3809,
+      "step": 813
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9777554003937414e-05,
+      "loss": 0.4475,
+      "step": 814
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.97766997508865e-05,
+      "loss": 0.4492,
+      "step": 815
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.977584387299623e-05,
+      "loss": 0.5127,
+      "step": 816
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9774986370360715e-05,
+      "loss": 0.3121,
+      "step": 817
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9774127243074248e-05,
+      "loss": 0.5253,
+      "step": 818
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.97732664912313e-05,
+      "loss": 0.4641,
+      "step": 819
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.977240411492652e-05,
+      "loss": 0.4639,
+      "step": 820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9771540114254723e-05,
+      "loss": 0.5203,
+      "step": 821
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9770674489310925e-05,
+      "loss": 0.457,
+      "step": 822
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9769807240190307e-05,
+      "loss": 0.3497,
+      "step": 823
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9768938366988232e-05,
+      "loss": 0.4978,
+      "step": 824
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.976806786980024e-05,
+      "loss": 0.374,
+      "step": 825
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9767195748722053e-05,
+      "loss": 0.5039,
+      "step": 826
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9766322003849565e-05,
+      "loss": 0.3896,
+      "step": 827
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.976544663527885e-05,
+      "loss": 0.3916,
+      "step": 828
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9764569643106172e-05,
+      "loss": 0.4923,
+      "step": 829
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.976369102742796e-05,
+      "loss": 0.3638,
+      "step": 830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9762810788340828e-05,
+      "loss": 0.4196,
+      "step": 831
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9761928925941558e-05,
+      "loss": 0.427,
+      "step": 832
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9761045440327128e-05,
+      "loss": 0.4707,
+      "step": 833
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9760160331594687e-05,
+      "loss": 0.3727,
+      "step": 834
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9759273599841558e-05,
+      "loss": 0.3623,
+      "step": 835
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9758385245165242e-05,
+      "loss": 0.4542,
+      "step": 836
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9757495267663432e-05,
+      "loss": 0.4855,
+      "step": 837
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9756603667433977e-05,
+      "loss": 0.4419,
+      "step": 838
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9755710444574927e-05,
+      "loss": 0.3938,
+      "step": 839
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9754815599184496e-05,
+      "loss": 0.439,
+      "step": 840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9753919131361083e-05,
+      "loss": 0.3918,
+      "step": 841
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.975302104120326e-05,
+      "loss": 0.4528,
+      "step": 842
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.975212132880978e-05,
+      "loss": 0.4712,
+      "step": 843
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.975121999427958e-05,
+      "loss": 0.4382,
+      "step": 844
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9750317037711765e-05,
+      "loss": 0.5166,
+      "step": 845
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9749412459205624e-05,
+      "loss": 0.3864,
+      "step": 846
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9748506258860632e-05,
+      "loss": 0.3693,
+      "step": 847
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.974759843677643e-05,
+      "loss": 0.4375,
+      "step": 848
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9746688993052833e-05,
+      "loss": 0.3208,
+      "step": 849
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9745777927789854e-05,
+      "loss": 0.3848,
+      "step": 850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.974486524108767e-05,
+      "loss": 0.3557,
+      "step": 851
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.974395093304664e-05,
+      "loss": 0.4019,
+      "step": 852
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.97430350037673e-05,
+      "loss": 0.3704,
+      "step": 853
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9742117453350364e-05,
+      "loss": 0.3577,
+      "step": 854
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.974119828189673e-05,
+      "loss": 0.4719,
+      "step": 855
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9740277489507463e-05,
+      "loss": 0.4517,
+      "step": 856
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.973935507628382e-05,
+      "loss": 0.4302,
+      "step": 857
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.973843104232722e-05,
+      "loss": 0.4858,
+      "step": 858
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.973750538773928e-05,
+      "loss": 0.439,
+      "step": 859
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9736578112621777e-05,
+      "loss": 0.4045,
+      "step": 860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.973564921707668e-05,
+      "loss": 0.4985,
+      "step": 861
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9734718701206126e-05,
+      "loss": 0.4741,
+      "step": 862
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.973378656511243e-05,
+      "loss": 0.4952,
+      "step": 863
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9732852808898097e-05,
+      "loss": 0.4468,
+      "step": 864
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9731917432665794e-05,
+      "loss": 0.4849,
+      "step": 865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9730980436518384e-05,
+      "loss": 0.3555,
+      "step": 866
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9730041820558896e-05,
+      "loss": 0.3691,
+      "step": 867
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9729101584890537e-05,
+      "loss": 0.3998,
+      "step": 868
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9728159729616694e-05,
+      "loss": 0.3333,
+      "step": 869
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9727216254840934e-05,
+      "loss": 0.4729,
+      "step": 870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9726271160667006e-05,
+      "loss": 0.4908,
+      "step": 871
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9725324447198824e-05,
+      "loss": 0.5193,
+      "step": 872
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.97243761145405e-05,
+      "loss": 0.4476,
+      "step": 873
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9723426162796295e-05,
+      "loss": 0.4277,
+      "step": 874
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.972247459207068e-05,
+      "loss": 0.4397,
+      "step": 875
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9721521402468285e-05,
+      "loss": 0.4592,
+      "step": 876
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9720566594093922e-05,
+      "loss": 0.366,
+      "step": 877
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9719610167052576e-05,
+      "loss": 0.318,
+      "step": 878
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9718652121449426e-05,
+      "loss": 0.3923,
+      "step": 879
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9717692457389806e-05,
+      "loss": 0.424,
+      "step": 880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.971673117497925e-05,
+      "loss": 0.5277,
+      "step": 881
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9715768274323458e-05,
+      "loss": 0.4554,
+      "step": 882
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9714803755528307e-05,
+      "loss": 0.4148,
+      "step": 883
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.971383761869986e-05,
+      "loss": 0.3687,
+      "step": 884
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9712869863944346e-05,
+      "loss": 0.4675,
+      "step": 885
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9711900491368186e-05,
+      "loss": 0.4237,
+      "step": 886
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9710929501077966e-05,
+      "loss": 0.3345,
+      "step": 887
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.970995689318046e-05,
+      "loss": 0.4272,
+      "step": 888
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9708982667782612e-05,
+      "loss": 0.488,
+      "step": 889
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.970800682499155e-05,
+      "loss": 0.3894,
+      "step": 890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.970702936491458e-05,
+      "loss": 0.395,
+      "step": 891
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9706050287659175e-05,
+      "loss": 0.3729,
+      "step": 892
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9705069593333003e-05,
+      "loss": 0.4092,
+      "step": 893
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9704087282043893e-05,
+      "loss": 0.3018,
+      "step": 894
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9703103353899857e-05,
+      "loss": 0.4304,
+      "step": 895
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.97021178090091e-05,
+      "loss": 0.4343,
+      "step": 896
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.970113064747999e-05,
+      "loss": 0.5212,
+      "step": 897
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9700141869421057e-05,
+      "loss": 0.3521,
+      "step": 898
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.969915147494105e-05,
+      "loss": 0.4441,
+      "step": 899
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.969815946414886e-05,
+      "loss": 0.4257,
+      "step": 900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9697165837153563e-05,
+      "loss": 0.4772,
+      "step": 901
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9696170594064433e-05,
+      "loss": 0.3776,
+      "step": 902
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9695173734990894e-05,
+      "loss": 0.3964,
+      "step": 903
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9694175260042567e-05,
+      "loss": 0.4346,
+      "step": 904
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9693175169329236e-05,
+      "loss": 0.3799,
+      "step": 905
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9692173462960883e-05,
+      "loss": 0.3778,
+      "step": 906
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.969117014104764e-05,
+      "loss": 0.4465,
+      "step": 907
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.969016520369985e-05,
+      "loss": 0.3325,
+      "step": 908
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9689158651027997e-05,
+      "loss": 0.5442,
+      "step": 909
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.968815048314277e-05,
+      "loss": 0.4741,
+      "step": 910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9687140700155027e-05,
+      "loss": 0.4608,
+      "step": 911
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9686129302175802e-05,
+      "loss": 0.3976,
+      "step": 912
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9685116289316312e-05,
+      "loss": 0.5652,
+      "step": 913
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9684101661687937e-05,
+      "loss": 0.5486,
+      "step": 914
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9683085419402258e-05,
+      "loss": 0.4341,
+      "step": 915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.968206756257101e-05,
+      "loss": 0.3966,
+      "step": 916
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.968104809130612e-05,
+      "loss": 0.3429,
+      "step": 917
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.968002700571969e-05,
+      "loss": 0.5103,
+      "step": 918
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9679004305924e-05,
+      "loss": 0.3632,
+      "step": 919
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9677979992031502e-05,
+      "loss": 0.3424,
+      "step": 920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9676954064154826e-05,
+      "loss": 0.468,
+      "step": 921
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9675926522406787e-05,
+      "loss": 0.4301,
+      "step": 922
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9674897366900376e-05,
+      "loss": 0.3997,
+      "step": 923
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.967386659774875e-05,
+      "loss": 0.4575,
+      "step": 924
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.967283421506526e-05,
+      "loss": 0.354,
+      "step": 925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9671800218963424e-05,
+      "loss": 0.2888,
+      "step": 926
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9670764609556932e-05,
+      "loss": 0.5356,
+      "step": 927
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9669727386959676e-05,
+      "loss": 0.4031,
+      "step": 928
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.966868855128569e-05,
+      "loss": 0.4534,
+      "step": 929
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9667648102649214e-05,
+      "loss": 0.4414,
+      "step": 930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9666606041164657e-05,
+      "loss": 0.498,
+      "step": 931
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9665562366946598e-05,
+      "loss": 0.4495,
+      "step": 932
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.96645170801098e-05,
+      "loss": 0.3921,
+      "step": 933
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.966347018076921e-05,
+      "loss": 0.3517,
+      "step": 934
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.966242166903993e-05,
+      "loss": 0.4502,
+      "step": 935
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.966137154503727e-05,
+      "loss": 0.4045,
+      "step": 936
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9660319808876684e-05,
+      "loss": 0.3898,
+      "step": 937
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9659266460673837e-05,
+      "loss": 0.4597,
+      "step": 938
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9658211500544542e-05,
+      "loss": 0.5,
+      "step": 939
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.965715492860481e-05,
+      "loss": 0.407,
+      "step": 940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.965609674497082e-05,
+      "loss": 0.4109,
+      "step": 941
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.965503694975893e-05,
+      "loss": 0.4565,
+      "step": 942
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9653975543085666e-05,
+      "loss": 0.3417,
+      "step": 943
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9652912525067746e-05,
+      "loss": 0.3484,
+      "step": 944
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9651847895822067e-05,
+      "loss": 0.3658,
+      "step": 945
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.965078165546568e-05,
+      "loss": 0.3937,
+      "step": 946
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.964971380411584e-05,
+      "loss": 0.3982,
+      "step": 947
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9648644341889965e-05,
+      "loss": 0.3859,
+      "step": 948
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9647573268905646e-05,
+      "loss": 0.3464,
+      "step": 949
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.964650058528066e-05,
+      "loss": 0.3956,
+      "step": 950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9645426291132966e-05,
+      "loss": 0.3981,
+      "step": 951
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.964435038658069e-05,
+      "loss": 0.3401,
+      "step": 952
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9643272871742133e-05,
+      "loss": 0.3298,
+      "step": 953
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9642193746735782e-05,
+      "loss": 0.4375,
+      "step": 954
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9641113011680295e-05,
+      "loss": 0.4014,
+      "step": 955
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.964003066669451e-05,
+      "loss": 0.4617,
+      "step": 956
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9638946711897447e-05,
+      "loss": 0.3484,
+      "step": 957
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.963786114740829e-05,
+      "loss": 0.4886,
+      "step": 958
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9636773973346407e-05,
+      "loss": 0.3574,
+      "step": 959
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9635685189831343e-05,
+      "loss": 0.4375,
+      "step": 960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9634594796982827e-05,
+      "loss": 0.339,
+      "step": 961
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9633502794920752e-05,
+      "loss": 0.3794,
+      "step": 962
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9632409183765196e-05,
+      "loss": 0.4144,
+      "step": 963
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9631313963636412e-05,
+      "loss": 0.4023,
+      "step": 964
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9630217134654826e-05,
+      "loss": 0.3899,
+      "step": 965
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.962911869694105e-05,
+      "loss": 0.4186,
+      "step": 966
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9628018650615867e-05,
+      "loss": 0.3757,
+      "step": 967
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9626916995800237e-05,
+      "loss": 0.3851,
+      "step": 968
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.96258137326153e-05,
+      "loss": 0.36,
+      "step": 969
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.962470886118236e-05,
+      "loss": 0.4172,
+      "step": 970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9623602381622923e-05,
+      "loss": 0.3945,
+      "step": 971
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9622494294058645e-05,
+      "loss": 0.5645,
+      "step": 972
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9621384598611377e-05,
+      "loss": 0.4321,
+      "step": 973
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9620273295403138e-05,
+      "loss": 0.3768,
+      "step": 974
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9619160384556133e-05,
+      "loss": 0.4019,
+      "step": 975
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9618045866192724e-05,
+      "loss": 0.4009,
+      "step": 976
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9616929740435474e-05,
+      "loss": 0.4034,
+      "step": 977
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.961581200740711e-05,
+      "loss": 0.3851,
+      "step": 978
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.961469266723053e-05,
+      "loss": 0.3683,
+      "step": 979
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.961357172002883e-05,
+      "loss": 0.3738,
+      "step": 980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9612449165925256e-05,
+      "loss": 0.4932,
+      "step": 981
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.961132500504325e-05,
+      "loss": 0.3572,
+      "step": 982
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9610199237506416e-05,
+      "loss": 0.4514,
+      "step": 983
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9609071863438557e-05,
+      "loss": 0.3788,
+      "step": 984
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.960794288296363e-05,
+      "loss": 0.4574,
+      "step": 985
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9606812296205777e-05,
+      "loss": 0.4875,
+      "step": 986
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9605680103289317e-05,
+      "loss": 0.4521,
+      "step": 987
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9604546304338748e-05,
+      "loss": 0.3337,
+      "step": 988
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.960341089947874e-05,
+      "loss": 0.4712,
+      "step": 989
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.960227388883414e-05,
+      "loss": 0.4741,
+      "step": 990
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.960113527252998e-05,
+      "loss": 0.3516,
+      "step": 991
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9599995050691453e-05,
+      "loss": 0.4116,
+      "step": 992
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.959885322344394e-05,
+      "loss": 0.3879,
+      "step": 993
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9597709790913004e-05,
+      "loss": 0.3811,
+      "step": 994
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9596564753224366e-05,
+      "loss": 0.4066,
+      "step": 995
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9595418110503932e-05,
+      "loss": 0.4095,
+      "step": 996
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9594269862877797e-05,
+      "loss": 0.4136,
+      "step": 997
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9593120010472215e-05,
+      "loss": 0.3365,
+      "step": 998
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9591968553413624e-05,
+      "loss": 0.446,
+      "step": 999
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.959081549182864e-05,
+      "loss": 0.3372,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9589660825844052e-05,
+      "loss": 0.4526,
+      "step": 1001
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9588504555586824e-05,
+      "loss": 0.4027,
+      "step": 1002
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.95873466811841e-05,
+      "loss": 0.3885,
+      "step": 1003
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9586187202763203e-05,
+      "loss": 0.3939,
+      "step": 1004
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.958502612045163e-05,
+      "loss": 0.344,
+      "step": 1005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9583863434377044e-05,
+      "loss": 0.4818,
+      "step": 1006
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9582699144667296e-05,
+      "loss": 0.366,
+      "step": 1007
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9581533251450416e-05,
+      "loss": 0.3368,
+      "step": 1008
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9580365754854604e-05,
+      "loss": 0.3264,
+      "step": 1009
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.957919665500824e-05,
+      "loss": 0.4109,
+      "step": 1010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9578025952039864e-05,
+      "loss": 0.3469,
+      "step": 1011
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9576853646078222e-05,
+      "loss": 0.3776,
+      "step": 1012
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9575679737252213e-05,
+      "loss": 0.3585,
+      "step": 1013
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9574504225690923e-05,
+      "loss": 0.4108,
+      "step": 1014
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.95733271115236e-05,
+      "loss": 0.3508,
+      "step": 1015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9572148394879695e-05,
+      "loss": 0.3649,
+      "step": 1016
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9570968075888807e-05,
+      "loss": 0.4175,
+      "step": 1017
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.956978615468073e-05,
+      "loss": 0.437,
+      "step": 1018
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9568602631385425e-05,
+      "loss": 0.2887,
+      "step": 1019
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9567417506133032e-05,
+      "loss": 0.5378,
+      "step": 1020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.956623077905386e-05,
+      "loss": 0.4431,
+      "step": 1021
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.956504245027842e-05,
+      "loss": 0.3862,
+      "step": 1022
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.956385251993736e-05,
+      "loss": 0.3162,
+      "step": 1023
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.956266098816153e-05,
+      "loss": 0.4186,
+      "step": 1024
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9561467855081957e-05,
+      "loss": 0.4089,
+      "step": 1025
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9560273120829828e-05,
+      "loss": 0.3846,
+      "step": 1026
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9559076785536526e-05,
+      "loss": 0.39,
+      "step": 1027
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.955787884933359e-05,
+      "loss": 0.3102,
+      "step": 1028
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9556679312352744e-05,
+      "loss": 0.4517,
+      "step": 1029
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.95554781747259e-05,
+      "loss": 0.375,
+      "step": 1030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9554275436585126e-05,
+      "loss": 0.3853,
+      "step": 1031
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9553071098062675e-05,
+      "loss": 0.4619,
+      "step": 1032
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9551865159290975e-05,
+      "loss": 0.3466,
+      "step": 1033
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9550657620402632e-05,
+      "loss": 0.3572,
+      "step": 1034
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9549448481530428e-05,
+      "loss": 0.4086,
+      "step": 1035
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9548237742807316e-05,
+      "loss": 0.4095,
+      "step": 1036
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.954702540436643e-05,
+      "loss": 0.3691,
+      "step": 1037
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9545811466341078e-05,
+      "loss": 0.3381,
+      "step": 1038
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9544595928864747e-05,
+      "loss": 0.4336,
+      "step": 1039
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.954337879207109e-05,
+      "loss": 0.3462,
+      "step": 1040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.954216005609395e-05,
+      "loss": 0.5046,
+      "step": 1041
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9540939721067334e-05,
+      "loss": 0.4773,
+      "step": 1042
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.953971778712543e-05,
+      "loss": 0.4572,
+      "step": 1043
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9538494254402604e-05,
+      "loss": 0.4363,
+      "step": 1044
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9537269123033397e-05,
+      "loss": 0.3909,
+      "step": 1045
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9536042393152512e-05,
+      "loss": 0.3951,
+      "step": 1046
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9534814064894855e-05,
+      "loss": 0.46,
+      "step": 1047
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9533584138395488e-05,
+      "loss": 0.4148,
+      "step": 1048
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9532352613789646e-05,
+      "loss": 0.3813,
+      "step": 1049
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.953111949121275e-05,
+      "loss": 0.3865,
+      "step": 1050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9529884770800398e-05,
+      "loss": 0.5117,
+      "step": 1051
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9528648452688358e-05,
+      "loss": 0.3323,
+      "step": 1052
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.952741053701257e-05,
+      "loss": 0.3982,
+      "step": 1053
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.952617102390916e-05,
+      "loss": 0.4929,
+      "step": 1054
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9524929913514424e-05,
+      "loss": 0.3301,
+      "step": 1055
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.952368720596483e-05,
+      "loss": 0.3944,
+      "step": 1056
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.952244290139703e-05,
+      "loss": 0.4102,
+      "step": 1057
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9521196999947846e-05,
+      "loss": 0.3918,
+      "step": 1058
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9519949501754273e-05,
+      "loss": 0.3701,
+      "step": 1059
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.951870040695349e-05,
+      "loss": 0.4128,
+      "step": 1060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9517449715682847e-05,
+      "loss": 0.377,
+      "step": 1061
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9516197428079866e-05,
+      "loss": 0.3126,
+      "step": 1062
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9514943544282245e-05,
+      "loss": 0.4521,
+      "step": 1063
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9513688064427876e-05,
+      "loss": 0.4205,
+      "step": 1064
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9512430988654794e-05,
+      "loss": 0.3945,
+      "step": 1065
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.951117231710123e-05,
+      "loss": 0.3643,
+      "step": 1066
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9509912049905595e-05,
+      "loss": 0.437,
+      "step": 1067
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9508650187206456e-05,
+      "loss": 0.4009,
+      "step": 1068
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.950738672914258e-05,
+      "loss": 0.3943,
+      "step": 1069
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.950612167585288e-05,
+      "loss": 0.3994,
+      "step": 1070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9504855027476478e-05,
+      "loss": 0.4316,
+      "step": 1071
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9503586784152645e-05,
+      "loss": 0.4248,
+      "step": 1072
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9502316946020837e-05,
+      "loss": 0.3334,
+      "step": 1073
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.950104551322069e-05,
+      "loss": 0.3737,
+      "step": 1074
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9499772485891995e-05,
+      "loss": 0.4504,
+      "step": 1075
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.949849786417475e-05,
+      "loss": 0.2689,
+      "step": 1076
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.949722164820911e-05,
+      "loss": 0.4553,
+      "step": 1077
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.94959438381354e-05,
+      "loss": 0.3007,
+      "step": 1078
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.949466443409413e-05,
+      "loss": 0.3812,
+      "step": 1079
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.949338343622598e-05,
+      "loss": 0.34,
+      "step": 1080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.949210084467181e-05,
+      "loss": 0.406,
+      "step": 1081
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.949081665957266e-05,
+      "loss": 0.4402,
+      "step": 1082
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9489530881069728e-05,
+      "loss": 0.4249,
+      "step": 1083
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.948824350930441e-05,
+      "loss": 0.4559,
+      "step": 1084
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.948695454441825e-05,
+      "loss": 0.4333,
+      "step": 1085
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9485663986552992e-05,
+      "loss": 0.3789,
+      "step": 1086
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.948437183585054e-05,
+      "loss": 0.3297,
+      "step": 1087
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9483078092452984e-05,
+      "loss": 0.4446,
+      "step": 1088
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.948178275650258e-05,
+      "loss": 0.4176,
+      "step": 1089
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.948048582814176e-05,
+      "loss": 0.3777,
+      "step": 1090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.947918730751314e-05,
+      "loss": 0.2913,
+      "step": 1091
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.94778871947595e-05,
+      "loss": 0.5404,
+      "step": 1092
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.94765854900238e-05,
+      "loss": 0.3503,
+      "step": 1093
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9475282193449175e-05,
+      "loss": 0.3475,
+      "step": 1094
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.947397730517894e-05,
+      "loss": 0.3793,
+      "step": 1095
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9472670825356572e-05,
+      "loss": 0.3894,
+      "step": 1096
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.947136275412574e-05,
+      "loss": 0.4029,
+      "step": 1097
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9470053091630265e-05,
+      "loss": 0.3953,
+      "step": 1098
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.946874183801417e-05,
+      "loss": 0.4624,
+      "step": 1099
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9467428993421635e-05,
+      "loss": 0.3992,
+      "step": 1100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9466114557997022e-05,
+      "loss": 0.3469,
+      "step": 1101
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9464798531884863e-05,
+      "loss": 0.3757,
+      "step": 1102
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9463480915229866e-05,
+      "loss": 0.3549,
+      "step": 1103
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9462161708176922e-05,
+      "loss": 0.365,
+      "step": 1104
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9460840910871083e-05,
+      "loss": 0.366,
+      "step": 1105
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9459518523457587e-05,
+      "loss": 0.4425,
+      "step": 1106
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9458194546081845e-05,
+      "loss": 0.3812,
+      "step": 1107
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.945686897888944e-05,
+      "loss": 0.4126,
+      "step": 1108
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9455541822026127e-05,
+      "loss": 0.4143,
+      "step": 1109
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9454213075637842e-05,
+      "loss": 0.3552,
+      "step": 1110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9452882739870694e-05,
+      "loss": 0.3352,
+      "step": 1111
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9451550814870968e-05,
+      "loss": 0.3984,
+      "step": 1112
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.945021730078512e-05,
+      "loss": 0.3217,
+      "step": 1113
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.944888219775978e-05,
+      "loss": 0.2759,
+      "step": 1114
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.944754550594176e-05,
+      "loss": 0.4025,
+      "step": 1115
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9446207225478035e-05,
+      "loss": 0.3905,
+      "step": 1116
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.944486735651577e-05,
+      "loss": 0.3596,
+      "step": 1117
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.944352589920229e-05,
+      "loss": 0.4742,
+      "step": 1118
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.944218285368511e-05,
+      "loss": 0.3451,
+      "step": 1119
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9440838220111898e-05,
+      "loss": 0.3682,
+      "step": 1120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.943949199863052e-05,
+      "loss": 0.3368,
+      "step": 1121
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9438144189388997e-05,
+      "loss": 0.4124,
+      "step": 1122
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.943679479253554e-05,
+      "loss": 0.3912,
+      "step": 1123
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.943544380821853e-05,
+      "loss": 0.5085,
+      "step": 1124
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9434091236586516e-05,
+      "loss": 0.3994,
+      "step": 1125
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9432737077788228e-05,
+      "loss": 0.3115,
+      "step": 1126
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9431381331972563e-05,
+      "loss": 0.4226,
+      "step": 1127
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9430023999288608e-05,
+      "loss": 0.448,
+      "step": 1128
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9428665079885608e-05,
+      "loss": 0.3387,
+      "step": 1129
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9427304573912987e-05,
+      "loss": 0.3914,
+      "step": 1130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9425942481520355e-05,
+      "loss": 0.3445,
+      "step": 1131
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.942457880285748e-05,
+      "loss": 0.3164,
+      "step": 1132
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9423213538074314e-05,
+      "loss": 0.4235,
+      "step": 1133
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9421846687320978e-05,
+      "loss": 0.3911,
+      "step": 1134
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9420478250747774e-05,
+      "loss": 0.4376,
+      "step": 1135
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.941910822850517e-05,
+      "loss": 0.3594,
+      "step": 1136
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.941773662074382e-05,
+      "loss": 0.3184,
+      "step": 1137
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.941636342761454e-05,
+      "loss": 0.4095,
+      "step": 1138
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9414988649268324e-05,
+      "loss": 0.4597,
+      "step": 1139
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.941361228585635e-05,
+      "loss": 0.3612,
+      "step": 1140
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9412234337529957e-05,
+      "loss": 0.4409,
+      "step": 1141
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9410854804440662e-05,
+      "loss": 0.4595,
+      "step": 1142
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.940947368674016e-05,
+      "loss": 0.4102,
+      "step": 1143
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9408090984580323e-05,
+      "loss": 0.3711,
+      "step": 1144
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9406706698113182e-05,
+      "loss": 0.4517,
+      "step": 1145
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.940532082749096e-05,
+      "loss": 0.3302,
+      "step": 1146
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9403933372866043e-05,
+      "loss": 0.3839,
+      "step": 1147
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9402544334390995e-05,
+      "loss": 0.4707,
+      "step": 1148
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.940115371221856e-05,
+      "loss": 0.3707,
+      "step": 1149
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9399761506501643e-05,
+      "loss": 0.4471,
+      "step": 1150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.939836771739333e-05,
+      "loss": 0.353,
+      "step": 1151
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.939697234504689e-05,
+      "loss": 0.3575,
+      "step": 1152
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.939557538961575e-05,
+      "loss": 0.3678,
+      "step": 1153
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.939417685125352e-05,
+      "loss": 0.3988,
+      "step": 1154
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9392776730113986e-05,
+      "loss": 0.3621,
+      "step": 1155
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.93913750263511e-05,
+      "loss": 0.3735,
+      "step": 1156
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9389971740118994e-05,
+      "loss": 0.3719,
+      "step": 1157
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.938856687157197e-05,
+      "loss": 0.4158,
+      "step": 1158
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9387160420864518e-05,
+      "loss": 0.4236,
+      "step": 1159
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9385752388151277e-05,
+      "loss": 0.3198,
+      "step": 1160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9384342773587084e-05,
+      "loss": 0.3278,
+      "step": 1161
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9382931577326934e-05,
+      "loss": 0.3654,
+      "step": 1162
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9381518799526002e-05,
+      "loss": 0.309,
+      "step": 1163
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9380104440339633e-05,
+      "loss": 0.4285,
+      "step": 1164
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9378688499923356e-05,
+      "loss": 0.3391,
+      "step": 1165
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.937727097843287e-05,
+      "loss": 0.5264,
+      "step": 1166
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9375851876024033e-05,
+      "loss": 0.4647,
+      "step": 1167
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9374431192852898e-05,
+      "loss": 0.4623,
+      "step": 1168
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9373008929075674e-05,
+      "loss": 0.3099,
+      "step": 1169
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9371585084848765e-05,
+      "loss": 0.2878,
+      "step": 1170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9370159660328732e-05,
+      "loss": 0.441,
+      "step": 1171
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9368732655672307e-05,
+      "loss": 0.3335,
+      "step": 1172
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.936730407103641e-05,
+      "loss": 0.3159,
+      "step": 1173
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9365873906578126e-05,
+      "loss": 0.3185,
+      "step": 1174
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9364442162454713e-05,
+      "loss": 0.4148,
+      "step": 1175
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9363008838823607e-05,
+      "loss": 0.276,
+      "step": 1176
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9361573935842417e-05,
+      "loss": 0.256,
+      "step": 1177
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9360137453668917e-05,
+      "loss": 0.366,
+      "step": 1178
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.935869939246107e-05,
+      "loss": 0.4583,
+      "step": 1179
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9357259752377007e-05,
+      "loss": 0.3324,
+      "step": 1180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.935581853357502e-05,
+      "loss": 0.3577,
+      "step": 1181
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9354375736213587e-05,
+      "loss": 0.4088,
+      "step": 1182
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9352931360451367e-05,
+      "loss": 0.355,
+      "step": 1183
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.935148540644717e-05,
+      "loss": 0.3883,
+      "step": 1184
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9350037874360005e-05,
+      "loss": 0.399,
+      "step": 1185
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.934858876434903e-05,
+      "loss": 0.3594,
+      "step": 1186
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9347138076573597e-05,
+      "loss": 0.3491,
+      "step": 1187
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9345685811193223e-05,
+      "loss": 0.3885,
+      "step": 1188
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.934423196836759e-05,
+      "loss": 0.371,
+      "step": 1189
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9342776548256567e-05,
+      "loss": 0.2893,
+      "step": 1190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.93413195510202e-05,
+      "loss": 0.4618,
+      "step": 1191
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.933986097681868e-05,
+      "loss": 0.232,
+      "step": 1192
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.933840082581241e-05,
+      "loss": 0.397,
+      "step": 1193
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.933693909816194e-05,
+      "loss": 0.2804,
+      "step": 1194
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9335475794028e-05,
+      "loss": 0.4753,
+      "step": 1195
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9334010913571493e-05,
+      "loss": 0.3961,
+      "step": 1196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9332544456953504e-05,
+      "loss": 0.436,
+      "step": 1197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.933107642433528e-05,
+      "loss": 0.3699,
+      "step": 1198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.932960681587824e-05,
+      "loss": 0.3361,
+      "step": 1199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9328135631743983e-05,
+      "loss": 0.4719,
+      "step": 1200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9326662872094288e-05,
+      "loss": 0.3877,
+      "step": 1201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9325188537091095e-05,
+      "loss": 0.3058,
+      "step": 1202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9323712626896516e-05,
+      "loss": 0.2979,
+      "step": 1203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9322235141672845e-05,
+      "loss": 0.3546,
+      "step": 1204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9320756081582546e-05,
+      "loss": 0.3502,
+      "step": 1205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9319275446788257e-05,
+      "loss": 0.343,
+      "step": 1206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9317793237452785e-05,
+      "loss": 0.3065,
+      "step": 1207
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9316309453739118e-05,
+      "loss": 0.337,
+      "step": 1208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9314824095810402e-05,
+      "loss": 0.3464,
+      "step": 1209
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.931333716382998e-05,
+      "loss": 0.3694,
+      "step": 1210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9311848657961344e-05,
+      "loss": 0.292,
+      "step": 1211
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9310358578368176e-05,
+      "loss": 0.3896,
+      "step": 1212
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9308866925214315e-05,
+      "loss": 0.3154,
+      "step": 1213
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9307373698663793e-05,
+      "loss": 0.3418,
+      "step": 1214
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9305878898880803e-05,
+      "loss": 0.3502,
+      "step": 1215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9304382526029708e-05,
+      "loss": 0.4087,
+      "step": 1216
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.930288458027505e-05,
+      "loss": 0.3674,
+      "step": 1217
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9301385061781542e-05,
+      "loss": 0.3625,
+      "step": 1218
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9299883970714073e-05,
+      "loss": 0.3983,
+      "step": 1219
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9298381307237697e-05,
+      "loss": 0.3627,
+      "step": 1220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9296877071517653e-05,
+      "loss": 0.3571,
+      "step": 1221
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9295371263719342e-05,
+      "loss": 0.4188,
+      "step": 1222
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9293863884008344e-05,
+      "loss": 0.3239,
+      "step": 1223
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.929235493255041e-05,
+      "loss": 0.3658,
+      "step": 1224
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.929084440951146e-05,
+      "loss": 0.4243,
+      "step": 1225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.928933231505759e-05,
+      "loss": 0.3445,
+      "step": 1226
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9287818649355078e-05,
+      "loss": 0.3756,
+      "step": 1227
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9286303412570358e-05,
+      "loss": 0.3812,
+      "step": 1228
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9284786604870046e-05,
+      "loss": 0.4429,
+      "step": 1229
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.928326822642093e-05,
+      "loss": 0.2894,
+      "step": 1230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.928174827738997e-05,
+      "loss": 0.3473,
+      "step": 1231
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.92802267579443e-05,
+      "loss": 0.3131,
+      "step": 1232
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9278703668251227e-05,
+      "loss": 0.3276,
+      "step": 1233
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9277179008478225e-05,
+      "loss": 0.4319,
+      "step": 1234
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9275652778792948e-05,
+      "loss": 0.3348,
+      "step": 1235
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9274124979363218e-05,
+      "loss": 0.4033,
+      "step": 1236
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9272595610357032e-05,
+      "loss": 0.3431,
+      "step": 1237
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9271064671942563e-05,
+      "loss": 0.35,
+      "step": 1238
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.926953216428814e-05,
+      "loss": 0.3408,
+      "step": 1239
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.926799808756229e-05,
+      "loss": 0.4246,
+      "step": 1240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9266462441933695e-05,
+      "loss": 0.3815,
+      "step": 1241
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.926492522757121e-05,
+      "loss": 0.3575,
+      "step": 1242
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.926338644464387e-05,
+      "loss": 0.3718,
+      "step": 1243
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.926184609332088e-05,
+      "loss": 0.2714,
+      "step": 1244
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9260304173771614e-05,
+      "loss": 0.3973,
+      "step": 1245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.925876068616562e-05,
+      "loss": 0.3043,
+      "step": 1246
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9257215630672624e-05,
+      "loss": 0.3175,
+      "step": 1247
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9255669007462513e-05,
+      "loss": 0.3911,
+      "step": 1248
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.925412081670536e-05,
+      "loss": 0.447,
+      "step": 1249
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9252571058571396e-05,
+      "loss": 0.329,
+      "step": 1250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.925101973323104e-05,
+      "loss": 0.3011,
+      "step": 1251
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9249466840854867e-05,
+      "loss": 0.2479,
+      "step": 1252
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9247912381613636e-05,
+      "loss": 0.3599,
+      "step": 1253
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9246356355678277e-05,
+      "loss": 0.3783,
+      "step": 1254
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9244798763219884e-05,
+      "loss": 0.3732,
+      "step": 1255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9243239604409733e-05,
+      "loss": 0.4491,
+      "step": 1256
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.924167887941927e-05,
+      "loss": 0.2911,
+      "step": 1257
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.924011658842011e-05,
+      "loss": 0.4027,
+      "step": 1258
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.923855273158404e-05,
+      "loss": 0.4182,
+      "step": 1259
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9236987309083023e-05,
+      "loss": 0.2858,
+      "step": 1260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9235420321089193e-05,
+      "loss": 0.3529,
+      "step": 1261
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9233851767774852e-05,
+      "loss": 0.3873,
+      "step": 1262
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9232281649312482e-05,
+      "loss": 0.3072,
+      "step": 1263
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.923070996587473e-05,
+      "loss": 0.3191,
+      "step": 1264
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9229136717634414e-05,
+      "loss": 0.4351,
+      "step": 1265
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.922756190476453e-05,
+      "loss": 0.3384,
+      "step": 1266
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.922598552743825e-05,
+      "loss": 0.3662,
+      "step": 1267
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9224407585828905e-05,
+      "loss": 0.4273,
+      "step": 1268
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9222828080110002e-05,
+      "loss": 0.4039,
+      "step": 1269
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9221247010455232e-05,
+      "loss": 0.3615,
+      "step": 1270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9219664377038443e-05,
+      "loss": 0.386,
+      "step": 1271
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9218080180033664e-05,
+      "loss": 0.35,
+      "step": 1272
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9216494419615088e-05,
+      "loss": 0.2819,
+      "step": 1273
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9214907095957086e-05,
+      "loss": 0.413,
+      "step": 1274
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.92133182092342e-05,
+      "loss": 0.2349,
+      "step": 1275
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9211727759621143e-05,
+      "loss": 0.2673,
+      "step": 1276
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.92101357472928e-05,
+      "loss": 0.3799,
+      "step": 1277
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9208542172424234e-05,
+      "loss": 0.333,
+      "step": 1278
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9206947035190663e-05,
+      "loss": 0.3854,
+      "step": 1279
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9205350335767496e-05,
+      "loss": 0.3129,
+      "step": 1280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.92037520743303e-05,
+      "loss": 0.3657,
+      "step": 1281
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.920215225105482e-05,
+      "loss": 0.4564,
+      "step": 1282
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.920055086611698e-05,
+      "loss": 0.322,
+      "step": 1283
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.919894791969285e-05,
+      "loss": 0.3936,
+      "step": 1284
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9197343411958706e-05,
+      "loss": 0.3276,
+      "step": 1285
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9195737343090973e-05,
+      "loss": 0.3934,
+      "step": 1286
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9194129713266257e-05,
+      "loss": 0.3345,
+      "step": 1287
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9192520522661324e-05,
+      "loss": 0.3234,
+      "step": 1288
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9190909771453126e-05,
+      "loss": 0.2899,
+      "step": 1289
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.918929745981878e-05,
+      "loss": 0.3367,
+      "step": 1290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9187683587935576e-05,
+      "loss": 0.481,
+      "step": 1291
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.918606815598097e-05,
+      "loss": 0.2811,
+      "step": 1292
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9184451164132597e-05,
+      "loss": 0.3527,
+      "step": 1293
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9182832612568263e-05,
+      "loss": 0.3391,
+      "step": 1294
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.918121250146594e-05,
+      "loss": 0.3524,
+      "step": 1295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9179590831003774e-05,
+      "loss": 0.2401,
+      "step": 1296
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9177967601360088e-05,
+      "loss": 0.3555,
+      "step": 1297
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9176342812713366e-05,
+      "loss": 0.3368,
+      "step": 1298
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9174716465242273e-05,
+      "loss": 0.3937,
+      "step": 1299
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9173088559125643e-05,
+      "loss": 0.4209,
+      "step": 1300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9171459094542475e-05,
+      "loss": 0.3387,
+      "step": 1301
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9169828071671943e-05,
+      "loss": 0.4119,
+      "step": 1302
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.91681954906934e-05,
+      "loss": 0.3817,
+      "step": 1303
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.916656135178636e-05,
+      "loss": 0.3286,
+      "step": 1304
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.916492565513051e-05,
+      "loss": 0.3607,
+      "step": 1305
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9163288400905718e-05,
+      "loss": 0.2741,
+      "step": 1306
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9161649589292002e-05,
+      "loss": 0.3632,
+      "step": 1307
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.916000922046958e-05,
+      "loss": 0.3391,
+      "step": 1308
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9158367294618823e-05,
+      "loss": 0.3708,
+      "step": 1309
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.915672381192027e-05,
+      "loss": 0.2182,
+      "step": 1310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9155078772554643e-05,
+      "loss": 0.4562,
+      "step": 1311
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9153432176702825e-05,
+      "loss": 0.3398,
+      "step": 1312
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.915178402454588e-05,
+      "loss": 0.3088,
+      "step": 1313
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9150134316265037e-05,
+      "loss": 0.4426,
+      "step": 1314
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9148483052041695e-05,
+      "loss": 0.3596,
+      "step": 1315
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.914683023205743e-05,
+      "loss": 0.3839,
+      "step": 1316
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9145175856493982e-05,
+      "loss": 0.415,
+      "step": 1317
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9143519925533262e-05,
+      "loss": 0.3471,
+      "step": 1318
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9141862439357364e-05,
+      "loss": 0.321,
+      "step": 1319
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.914020339814854e-05,
+      "loss": 0.3806,
+      "step": 1320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9138542802089217e-05,
+      "loss": 0.3188,
+      "step": 1321
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.913688065136199e-05,
+      "loss": 0.3801,
+      "step": 1322
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.913521694614964e-05,
+      "loss": 0.3134,
+      "step": 1323
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9133551686635096e-05,
+      "loss": 0.3617,
+      "step": 1324
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9131884873001467e-05,
+      "loss": 0.4277,
+      "step": 1325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9130216505432052e-05,
+      "loss": 0.3229,
+      "step": 1326
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9128546584110285e-05,
+      "loss": 0.2502,
+      "step": 1327
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9126875109219798e-05,
+      "loss": 0.459,
+      "step": 1328
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.912520208094439e-05,
+      "loss": 0.3224,
+      "step": 1329
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.912352749946802e-05,
+      "loss": 0.3232,
+      "step": 1330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9121851364974824e-05,
+      "loss": 0.3314,
+      "step": 1331
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9120173677649115e-05,
+      "loss": 0.3964,
+      "step": 1332
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9118494437675364e-05,
+      "loss": 0.3705,
+      "step": 1333
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.911681364523822e-05,
+      "loss": 0.3253,
+      "step": 1334
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9115131300522506e-05,
+      "loss": 0.3159,
+      "step": 1335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9113447403713216e-05,
+      "loss": 0.3732,
+      "step": 1336
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9111761954995497e-05,
+      "loss": 0.2449,
+      "step": 1337
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9110074954554696e-05,
+      "loss": 0.3518,
+      "step": 1338
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.91083864025763e-05,
+      "loss": 0.2844,
+      "step": 1339
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9106696299245986e-05,
+      "loss": 0.3854,
+      "step": 1340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.910500464474961e-05,
+      "loss": 0.2358,
+      "step": 1341
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9103311439273167e-05,
+      "loss": 0.4326,
+      "step": 1342
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.910161668300285e-05,
+      "loss": 0.3535,
+      "step": 1343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9099920376125015e-05,
+      "loss": 0.2953,
+      "step": 1344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9098222518826185e-05,
+      "loss": 0.4102,
+      "step": 1345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.909652311129305e-05,
+      "loss": 0.2663,
+      "step": 1346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9094822153712487e-05,
+      "loss": 0.2897,
+      "step": 1347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9093119646271525e-05,
+      "loss": 0.4072,
+      "step": 1348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9091415589157377e-05,
+      "loss": 0.3845,
+      "step": 1349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9089709982557414e-05,
+      "loss": 0.363,
+      "step": 1350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.908800282665918e-05,
+      "loss": 0.4093,
+      "step": 1351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.908629412165041e-05,
+      "loss": 0.2989,
+      "step": 1352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9084583867718975e-05,
+      "loss": 0.2741,
+      "step": 1353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9082872065052943e-05,
+      "loss": 0.3384,
+      "step": 1354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.908115871384054e-05,
+      "loss": 0.4413,
+      "step": 1355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9079443814270163e-05,
+      "loss": 0.3688,
+      "step": 1356
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.907772736653039e-05,
+      "loss": 0.4589,
+      "step": 1357
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9076009370809952e-05,
+      "loss": 0.3145,
+      "step": 1358
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9074289827297764e-05,
+      "loss": 0.3236,
+      "step": 1359
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9072568736182903e-05,
+      "loss": 0.3015,
+      "step": 1360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.907084609765462e-05,
+      "loss": 0.3475,
+      "step": 1361
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.906912191190234e-05,
+      "loss": 0.3816,
+      "step": 1362
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9067396179115652e-05,
+      "loss": 0.3954,
+      "step": 1363
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9065668899484308e-05,
+      "loss": 0.3301,
+      "step": 1364
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.906394007319825e-05,
+      "loss": 0.3427,
+      "step": 1365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9062209700447576e-05,
+      "loss": 0.3429,
+      "step": 1366
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9060477781422556e-05,
+      "loss": 0.3284,
+      "step": 1367
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.905874431631363e-05,
+      "loss": 0.337,
+      "step": 1368
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.905700930531141e-05,
+      "loss": 0.3094,
+      "step": 1369
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9055272748606678e-05,
+      "loss": 0.3516,
+      "step": 1370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9053534646390383e-05,
+      "loss": 0.3408,
+      "step": 1371
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9051794998853648e-05,
+      "loss": 0.2773,
+      "step": 1372
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9050053806187765e-05,
+      "loss": 0.3495,
+      "step": 1373
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.904831106858419e-05,
+      "loss": 0.3298,
+      "step": 1374
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9046566786234556e-05,
+      "loss": 0.2578,
+      "step": 1375
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9044820959330663e-05,
+      "loss": 0.3188,
+      "step": 1376
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9043073588064485e-05,
+      "loss": 0.3269,
+      "step": 1377
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.904132467262816e-05,
+      "loss": 0.3726,
+      "step": 1378
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9039574213213996e-05,
+      "loss": 0.3627,
+      "step": 1379
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9037822210014472e-05,
+      "loss": 0.3018,
+      "step": 1380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9036068663222245e-05,
+      "loss": 0.361,
+      "step": 1381
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9034313573030125e-05,
+      "loss": 0.322,
+      "step": 1382
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9032556939631108e-05,
+      "loss": 0.2942,
+      "step": 1383
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9030798763218347e-05,
+      "loss": 0.3864,
+      "step": 1384
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9029039043985174e-05,
+      "loss": 0.364,
+      "step": 1385
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9027277782125083e-05,
+      "loss": 0.3344,
+      "step": 1386
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9025514977831748e-05,
+      "loss": 0.3785,
+      "step": 1387
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9023750631299e-05,
+      "loss": 0.2971,
+      "step": 1388
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9021984742720847e-05,
+      "loss": 0.2799,
+      "step": 1389
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.902021731229147e-05,
+      "loss": 0.3197,
+      "step": 1390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.901844834020521e-05,
+      "loss": 0.3331,
+      "step": 1391
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9016677826656584e-05,
+      "loss": 0.3842,
+      "step": 1392
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9014905771840273e-05,
+      "loss": 0.2925,
+      "step": 1393
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.901313217595114e-05,
+      "loss": 0.283,
+      "step": 1394
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9011357039184205e-05,
+      "loss": 0.3495,
+      "step": 1395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9009580361734656e-05,
+      "loss": 0.3282,
+      "step": 1396
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9007802143797862e-05,
+      "loss": 0.3163,
+      "step": 1397
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9006022385569354e-05,
+      "loss": 0.3846,
+      "step": 1398
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.900424108724483e-05,
+      "loss": 0.3438,
+      "step": 1399
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9002458249020162e-05,
+      "loss": 0.3278,
+      "step": 1400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.900067387109139e-05,
+      "loss": 0.3356,
+      "step": 1401
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.899888795365473e-05,
+      "loss": 0.295,
+      "step": 1402
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8997100496906553e-05,
+      "loss": 0.275,
+      "step": 1403
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8995311501043407e-05,
+      "loss": 0.2896,
+      "step": 1404
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8993520966262014e-05,
+      "loss": 0.3373,
+      "step": 1405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8991728892759257e-05,
+      "loss": 0.3528,
+      "step": 1406
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8989935280732195e-05,
+      "loss": 0.4299,
+      "step": 1407
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8988140130378046e-05,
+      "loss": 0.3524,
+      "step": 1408
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.898634344189421e-05,
+      "loss": 0.3835,
+      "step": 1409
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8984545215478246e-05,
+      "loss": 0.4015,
+      "step": 1410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8982745451327892e-05,
+      "loss": 0.3286,
+      "step": 1411
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.898094414964104e-05,
+      "loss": 0.3062,
+      "step": 1412
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8979141310615775e-05,
+      "loss": 0.321,
+      "step": 1413
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.897733693445032e-05,
+      "loss": 0.2675,
+      "step": 1414
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8975531021343095e-05,
+      "loss": 0.4436,
+      "step": 1415
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.897372357149267e-05,
+      "loss": 0.3483,
+      "step": 1416
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8971914585097798e-05,
+      "loss": 0.3147,
+      "step": 1417
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.897010406235739e-05,
+      "loss": 0.4094,
+      "step": 1418
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8968292003470528e-05,
+      "loss": 0.4102,
+      "step": 1419
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.896647840863647e-05,
+      "loss": 0.3807,
+      "step": 1420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8964663278054638e-05,
+      "loss": 0.3429,
+      "step": 1421
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.896284661192462e-05,
+      "loss": 0.3068,
+      "step": 1422
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.896102841044618e-05,
+      "loss": 0.3387,
+      "step": 1423
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.895920867381924e-05,
+      "loss": 0.3241,
+      "step": 1424
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.89573874022439e-05,
+      "loss": 0.2911,
+      "step": 1425
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.895556459592043e-05,
+      "loss": 0.349,
+      "step": 1426
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8953740255049264e-05,
+      "loss": 0.2881,
+      "step": 1427
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8951914379831e-05,
+      "loss": 0.3331,
+      "step": 1428
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8950086970466418e-05,
+      "loss": 0.2482,
+      "step": 1429
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.894825802715646e-05,
+      "loss": 0.3875,
+      "step": 1430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8946427550102226e-05,
+      "loss": 0.3051,
+      "step": 1431
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8944595539505002e-05,
+      "loss": 0.3198,
+      "step": 1432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.894276199556623e-05,
+      "loss": 0.2883,
+      "step": 1433
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.894092691848753e-05,
+      "loss": 0.428,
+      "step": 1434
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8939090308470685e-05,
+      "loss": 0.3075,
+      "step": 1435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.893725216571765e-05,
+      "loss": 0.4122,
+      "step": 1436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8935412490430542e-05,
+      "loss": 0.2379,
+      "step": 1437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8933571282811654e-05,
+      "loss": 0.3359,
+      "step": 1438
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8931728543063444e-05,
+      "loss": 0.3947,
+      "step": 1439
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8929884271388535e-05,
+      "loss": 0.3662,
+      "step": 1440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8928038467989727e-05,
+      "loss": 0.3015,
+      "step": 1441
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8926191133069984e-05,
+      "loss": 0.2969,
+      "step": 1442
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8924342266832436e-05,
+      "loss": 0.3987,
+      "step": 1443
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8922491869480386e-05,
+      "loss": 0.3792,
+      "step": 1444
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8920639941217297e-05,
+      "loss": 0.3573,
+      "step": 1445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8918786482246807e-05,
+      "loss": 0.312,
+      "step": 1446
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8916931492772732e-05,
+      "loss": 0.3218,
+      "step": 1447
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8915074972999034e-05,
+      "loss": 0.3562,
+      "step": 1448
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.891321692312986e-05,
+      "loss": 0.3834,
+      "step": 1449
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8911357343369523e-05,
+      "loss": 0.275,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8909496233922494e-05,
+      "loss": 0.3696,
+      "step": 1451
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8907633594993426e-05,
+      "loss": 0.2592,
+      "step": 1452
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.890576942678713e-05,
+      "loss": 0.3331,
+      "step": 1453
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8903903729508597e-05,
+      "loss": 0.3531,
+      "step": 1454
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8902036503362967e-05,
+      "loss": 0.3737,
+      "step": 1455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8900167748555567e-05,
+      "loss": 0.2893,
+      "step": 1456
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.889829746529188e-05,
+      "loss": 0.438,
+      "step": 1457
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8896425653777568e-05,
+      "loss": 0.327,
+      "step": 1458
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.889455231421845e-05,
+      "loss": 0.3468,
+      "step": 1459
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8892677446820518e-05,
+      "loss": 0.3411,
+      "step": 1460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8890801051789923e-05,
+      "loss": 0.3958,
+      "step": 1461
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.888892312933301e-05,
+      "loss": 0.2911,
+      "step": 1462
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8887043679656263e-05,
+      "loss": 0.2853,
+      "step": 1463
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.888516270296635e-05,
+      "loss": 0.3745,
+      "step": 1464
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8883280199470096e-05,
+      "loss": 0.3341,
+      "step": 1465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.888139616937451e-05,
+      "loss": 0.304,
+      "step": 1466
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.887951061288675e-05,
+      "loss": 0.4298,
+      "step": 1467
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8877623530214157e-05,
+      "loss": 0.2772,
+      "step": 1468
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.887573492156423e-05,
+      "loss": 0.3918,
+      "step": 1469
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.887384478714464e-05,
+      "loss": 0.4082,
+      "step": 1470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8871953127163226e-05,
+      "loss": 0.2386,
+      "step": 1471
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8870059941828e-05,
+      "loss": 0.2908,
+      "step": 1472
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8868165231347124e-05,
+      "loss": 0.3871,
+      "step": 1473
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.886626899592895e-05,
+      "loss": 0.3896,
+      "step": 1474
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.886437123578198e-05,
+      "loss": 0.3358,
+      "step": 1475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8862471951114896e-05,
+      "loss": 0.4231,
+      "step": 1476
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8860571142136537e-05,
+      "loss": 0.2983,
+      "step": 1477
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8858668809055925e-05,
+      "loss": 0.2998,
+      "step": 1478
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.885676495208223e-05,
+      "loss": 0.2795,
+      "step": 1479
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8854859571424803e-05,
+      "loss": 0.3254,
+      "step": 1480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8852952667293157e-05,
+      "loss": 0.293,
+      "step": 1481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8851044239896972e-05,
+      "loss": 0.2816,
+      "step": 1482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.884913428944611e-05,
+      "loss": 0.3425,
+      "step": 1483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8847222816150574e-05,
+      "loss": 0.4126,
+      "step": 1484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8845309820220557e-05,
+      "loss": 0.2197,
+      "step": 1485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.884339530186641e-05,
+      "loss": 0.2972,
+      "step": 1486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.884147926129865e-05,
+      "loss": 0.3827,
+      "step": 1487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.883956169872796e-05,
+      "loss": 0.4116,
+      "step": 1488
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8837642614365205e-05,
+      "loss": 0.4457,
+      "step": 1489
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.88357220084214e-05,
+      "loss": 0.337,
+      "step": 1490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.883379988110774e-05,
+      "loss": 0.3527,
+      "step": 1491
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8831876232635568e-05,
+      "loss": 0.3234,
+      "step": 1492
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.882995106321642e-05,
+      "loss": 0.411,
+      "step": 1493
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8828024373061985e-05,
+      "loss": 0.2682,
+      "step": 1494
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8826096162384114e-05,
+      "loss": 0.3306,
+      "step": 1495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8824166431394844e-05,
+      "loss": 0.3266,
+      "step": 1496
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8822235180306354e-05,
+      "loss": 0.33,
+      "step": 1497
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8820302409331014e-05,
+      "loss": 0.3018,
+      "step": 1498
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.881836811868134e-05,
+      "loss": 0.3297,
+      "step": 1499
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.881643230857004e-05,
+      "loss": 0.3969,
+      "step": 1500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8814494979209966e-05,
+      "loss": 0.3692,
+      "step": 1501
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8812556130814144e-05,
+      "loss": 0.4114,
+      "step": 1502
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8810615763595773e-05,
+      "loss": 0.2663,
+      "step": 1503
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8808673877768216e-05,
+      "loss": 0.261,
+      "step": 1504
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8806730473544996e-05,
+      "loss": 0.213,
+      "step": 1505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8804785551139817e-05,
+      "loss": 0.2709,
+      "step": 1506
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8802839110766534e-05,
+      "loss": 0.3118,
+      "step": 1507
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8800891152639182e-05,
+      "loss": 0.3713,
+      "step": 1508
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8798941676971952e-05,
+      "loss": 0.2178,
+      "step": 1509
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8796990683979215e-05,
+      "loss": 0.329,
+      "step": 1510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8795038173875497e-05,
+      "loss": 0.3137,
+      "step": 1511
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8793084146875496e-05,
+      "loss": 0.2875,
+      "step": 1512
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8791128603194076e-05,
+      "loss": 0.2877,
+      "step": 1513
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8789171543046263e-05,
+      "loss": 0.3055,
+      "step": 1514
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8787212966647265e-05,
+      "loss": 0.3153,
+      "step": 1515
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8785252874212434e-05,
+      "loss": 0.3926,
+      "step": 1516
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8783291265957312e-05,
+      "loss": 0.2805,
+      "step": 1517
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8781328142097588e-05,
+      "loss": 0.3265,
+      "step": 1518
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.877936350284913e-05,
+      "loss": 0.3309,
+      "step": 1519
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.877739734842797e-05,
+      "loss": 0.2237,
+      "step": 1520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8775429679050307e-05,
+      "loss": 0.3493,
+      "step": 1521
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8773460494932497e-05,
+      "loss": 0.2802,
+      "step": 1522
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8771489796291077e-05,
+      "loss": 0.274,
+      "step": 1523
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.876951758334274e-05,
+      "loss": 0.2771,
+      "step": 1524
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8767543856304363e-05,
+      "loss": 0.3496,
+      "step": 1525
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8765568615392954e-05,
+      "loss": 0.428,
+      "step": 1526
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8763591860825726e-05,
+      "loss": 0.2892,
+      "step": 1527
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8761613592820038e-05,
+      "loss": 0.3837,
+      "step": 1528
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.875963381159342e-05,
+      "loss": 0.3358,
+      "step": 1529
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8757652517363563e-05,
+      "loss": 0.3635,
+      "step": 1530
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8755669710348337e-05,
+      "loss": 0.3401,
+      "step": 1531
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.875368539076576e-05,
+      "loss": 0.3157,
+      "step": 1532
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8751699558834038e-05,
+      "loss": 0.298,
+      "step": 1533
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.874971221477153e-05,
+      "loss": 0.2828,
+      "step": 1534
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.874772335879676e-05,
+      "loss": 0.2997,
+      "step": 1535
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8745732991128416e-05,
+      "loss": 0.3439,
+      "step": 1536
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.874374111198537e-05,
+      "loss": 0.3152,
+      "step": 1537
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8741747721586644e-05,
+      "loss": 0.3021,
+      "step": 1538
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8739752820151426e-05,
+      "loss": 0.3896,
+      "step": 1539
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.873775640789908e-05,
+      "loss": 0.3152,
+      "step": 1540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.873575848504913e-05,
+      "loss": 0.2745,
+      "step": 1541
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8733759051821263e-05,
+      "loss": 0.3252,
+      "step": 1542
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8731758108435335e-05,
+      "loss": 0.2304,
+      "step": 1543
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8729755655111376e-05,
+      "loss": 0.3037,
+      "step": 1544
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8727751692069566e-05,
+      "loss": 0.3093,
+      "step": 1545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.872574621953027e-05,
+      "loss": 0.313,
+      "step": 1546
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8723739237713997e-05,
+      "loss": 0.3018,
+      "step": 1547
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8721730746841447e-05,
+      "loss": 0.3711,
+      "step": 1548
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8719720747133458e-05,
+      "loss": 0.3065,
+      "step": 1549
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8717709238811062e-05,
+      "loss": 0.2632,
+      "step": 1550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8715696222095436e-05,
+      "loss": 0.3298,
+      "step": 1551
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8713681697207934e-05,
+      "loss": 0.3258,
+      "step": 1552
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.871166566437007e-05,
+      "loss": 0.2369,
+      "step": 1553
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.870964812380353e-05,
+      "loss": 0.1826,
+      "step": 1554
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.870762907573016e-05,
+      "loss": 0.3607,
+      "step": 1555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.870560852037196e-05,
+      "loss": 0.3182,
+      "step": 1556
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8703586457951134e-05,
+      "loss": 0.3606,
+      "step": 1557
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.870156288869001e-05,
+      "loss": 0.3499,
+      "step": 1558
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.86995378128111e-05,
+      "loss": 0.3175,
+      "step": 1559
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8697511230537094e-05,
+      "loss": 0.3218,
+      "step": 1560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.869548314209082e-05,
+      "loss": 0.2731,
+      "step": 1561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8693453547695285e-05,
+      "loss": 0.3911,
+      "step": 1562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8691422447573674e-05,
+      "loss": 0.323,
+      "step": 1563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8689389841949313e-05,
+      "loss": 0.3163,
+      "step": 1564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.868735573104572e-05,
+      "loss": 0.3601,
+      "step": 1565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8685320115086552e-05,
+      "loss": 0.3904,
+      "step": 1566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.868328299429565e-05,
+      "loss": 0.3488,
+      "step": 1567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8681244368897014e-05,
+      "loss": 0.3751,
+      "step": 1568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8679204239114815e-05,
+      "loss": 0.2351,
+      "step": 1569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8677162605173372e-05,
+      "loss": 0.3556,
+      "step": 1570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8675119467297202e-05,
+      "loss": 0.2471,
+      "step": 1571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8673074825710947e-05,
+      "loss": 0.325,
+      "step": 1572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.867102868063945e-05,
+      "loss": 0.2872,
+      "step": 1573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8668981032307698e-05,
+      "loss": 0.2948,
+      "step": 1574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.866693188094085e-05,
+      "loss": 0.2868,
+      "step": 1575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8664881226764227e-05,
+      "loss": 0.3564,
+      "step": 1576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8662829070003325e-05,
+      "loss": 0.2792,
+      "step": 1577
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8660775410883794e-05,
+      "loss": 0.2839,
+      "step": 1578
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.865872024963145e-05,
+      "loss": 0.3621,
+      "step": 1579
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8656663586472286e-05,
+      "loss": 0.3484,
+      "step": 1580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8654605421632444e-05,
+      "loss": 0.2983,
+      "step": 1581
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8652545755338243e-05,
+      "loss": 0.4432,
+      "step": 1582
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.865048458781616e-05,
+      "loss": 0.2697,
+      "step": 1583
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8648421919292843e-05,
+      "loss": 0.2713,
+      "step": 1584
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.86463577499951e-05,
+      "loss": 0.3508,
+      "step": 1585
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.864429208014991e-05,
+      "loss": 0.3835,
+      "step": 1586
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8642224909984406e-05,
+      "loss": 0.2686,
+      "step": 1587
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.86401562397259e-05,
+      "loss": 0.2277,
+      "step": 1588
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8638086069601864e-05,
+      "loss": 0.3809,
+      "step": 1589
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8636014399839924e-05,
+      "loss": 0.3181,
+      "step": 1590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8633941230667884e-05,
+      "loss": 0.3228,
+      "step": 1591
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.863186656231371e-05,
+      "loss": 0.319,
+      "step": 1592
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8629790395005532e-05,
+      "loss": 0.3876,
+      "step": 1593
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.862771272897164e-05,
+      "loss": 0.2834,
+      "step": 1594
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8625633564440495e-05,
+      "loss": 0.3279,
+      "step": 1595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.862355290164073e-05,
+      "loss": 0.3472,
+      "step": 1596
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.862147074080112e-05,
+      "loss": 0.2894,
+      "step": 1597
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8619387082150626e-05,
+      "loss": 0.3997,
+      "step": 1598
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8617301925918366e-05,
+      "loss": 0.4302,
+      "step": 1599
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8615215272333623e-05,
+      "loss": 0.327,
+      "step": 1600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.861312712162584e-05,
+      "loss": 0.3665,
+      "step": 1601
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8611037474024637e-05,
+      "loss": 0.3651,
+      "step": 1602
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.860894632975978e-05,
+      "loss": 0.3066,
+      "step": 1603
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.860685368906122e-05,
+      "loss": 0.4154,
+      "step": 1604
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.860475955215906e-05,
+      "loss": 0.2826,
+      "step": 1605
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.860266391928357e-05,
+      "loss": 0.2969,
+      "step": 1606
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8600566790665177e-05,
+      "loss": 0.361,
+      "step": 1607
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8598468166534494e-05,
+      "loss": 0.34,
+      "step": 1608
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.859636804712228e-05,
+      "loss": 0.2462,
+      "step": 1609
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.859426643265946e-05,
+      "loss": 0.2361,
+      "step": 1610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8592163323377128e-05,
+      "loss": 0.3245,
+      "step": 1611
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.859005871950654e-05,
+      "loss": 0.257,
+      "step": 1612
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8587952621279124e-05,
+      "loss": 0.2487,
+      "step": 1613
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8585845028926456e-05,
+      "loss": 0.2812,
+      "step": 1614
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.858373594268029e-05,
+      "loss": 0.3344,
+      "step": 1615
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.858162536277254e-05,
+      "loss": 0.2842,
+      "step": 1616
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8579513289435287e-05,
+      "loss": 0.3117,
+      "step": 1617
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.857739972290077e-05,
+      "loss": 0.2916,
+      "step": 1618
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8575284663401395e-05,
+      "loss": 0.2244,
+      "step": 1619
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.857316811116974e-05,
+      "loss": 0.2574,
+      "step": 1620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.857105006643853e-05,
+      "loss": 0.3121,
+      "step": 1621
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8568930529440677e-05,
+      "loss": 0.269,
+      "step": 1622
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.856680950040923e-05,
+      "loss": 0.2937,
+      "step": 1623
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.856468697957743e-05,
+      "loss": 0.3481,
+      "step": 1624
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8562562967178658e-05,
+      "loss": 0.2509,
+      "step": 1625
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8560437463446474e-05,
+      "loss": 0.3757,
+      "step": 1626
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8558310468614596e-05,
+      "loss": 0.2347,
+      "step": 1627
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8556181982916908e-05,
+      "loss": 0.29,
+      "step": 1628
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8554052006587457e-05,
+      "loss": 0.376,
+      "step": 1629
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.855192053986046e-05,
+      "loss": 0.3468,
+      "step": 1630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.854978758297028e-05,
+      "loss": 0.2833,
+      "step": 1631
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.854765313615147e-05,
+      "loss": 0.3116,
+      "step": 1632
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.854551719963872e-05,
+      "loss": 0.2906,
+      "step": 1633
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.85433797736669e-05,
+      "loss": 0.265,
+      "step": 1634
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8541240858471053e-05,
+      "loss": 0.3877,
+      "step": 1635
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8539100454286357e-05,
+      "loss": 0.2797,
+      "step": 1636
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8536958561348175e-05,
+      "loss": 0.3004,
+      "step": 1637
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.853481517989203e-05,
+      "loss": 0.2877,
+      "step": 1638
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8532670310153606e-05,
+      "loss": 0.3395,
+      "step": 1639
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8530523952368755e-05,
+      "loss": 0.2557,
+      "step": 1640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8528376106773487e-05,
+      "loss": 0.3779,
+      "step": 1641
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.852622677360398e-05,
+      "loss": 0.269,
+      "step": 1642
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8524075953096568e-05,
+      "loss": 0.2474,
+      "step": 1643
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.852192364548776e-05,
+      "loss": 0.2539,
+      "step": 1644
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8519769851014222e-05,
+      "loss": 0.2371,
+      "step": 1645
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8517614569912784e-05,
+      "loss": 0.2845,
+      "step": 1646
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8515457802420438e-05,
+      "loss": 0.3778,
+      "step": 1647
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8513299548774345e-05,
+      "loss": 0.3145,
+      "step": 1648
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.851113980921182e-05,
+      "loss": 0.2904,
+      "step": 1649
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.850897858397035e-05,
+      "loss": 0.2936,
+      "step": 1650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.850681587328758e-05,
+      "loss": 0.3629,
+      "step": 1651
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.850465167740133e-05,
+      "loss": 0.3892,
+      "step": 1652
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8502485996549562e-05,
+      "loss": 0.3036,
+      "step": 1653
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8500318830970422e-05,
+      "loss": 0.3425,
+      "step": 1654
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8498150180902203e-05,
+      "loss": 0.3036,
+      "step": 1655
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8495980046583375e-05,
+      "loss": 0.3785,
+      "step": 1656
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.849380842825256e-05,
+      "loss": 0.2466,
+      "step": 1657
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8491635326148558e-05,
+      "loss": 0.2595,
+      "step": 1658
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8489460740510307e-05,
+      "loss": 0.2994,
+      "step": 1659
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8487284671576937e-05,
+      "loss": 0.2919,
+      "step": 1660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8485107119587722e-05,
+      "loss": 0.3183,
+      "step": 1661
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8482928084782104e-05,
+      "loss": 0.2368,
+      "step": 1662
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8480747567399692e-05,
+      "loss": 0.3734,
+      "step": 1663
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.847856556768025e-05,
+      "loss": 0.2625,
+      "step": 1664
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.847638208586371e-05,
+      "loss": 0.3369,
+      "step": 1665
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8474197122190172e-05,
+      "loss": 0.2314,
+      "step": 1666
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.847201067689989e-05,
+      "loss": 0.4131,
+      "step": 1667
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8469822750233287e-05,
+      "loss": 0.4048,
+      "step": 1668
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.846763334243094e-05,
+      "loss": 0.293,
+      "step": 1669
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.84654424537336e-05,
+      "loss": 0.3764,
+      "step": 1670
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.846325008438218e-05,
+      "loss": 0.2884,
+      "step": 1671
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8461056234617746e-05,
+      "loss": 0.272,
+      "step": 1672
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.845886090468154e-05,
+      "loss": 0.2307,
+      "step": 1673
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8456664094814947e-05,
+      "loss": 0.2988,
+      "step": 1674
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8454465805259536e-05,
+      "loss": 0.2933,
+      "step": 1675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8452266036257027e-05,
+      "loss": 0.361,
+      "step": 1676
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.845006478804931e-05,
+      "loss": 0.2023,
+      "step": 1677
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.844786206087843e-05,
+      "loss": 0.2218,
+      "step": 1678
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8445657854986597e-05,
+      "loss": 0.239,
+      "step": 1679
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8443452170616188e-05,
+      "loss": 0.3176,
+      "step": 1680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8441245008009734e-05,
+      "loss": 0.1978,
+      "step": 1681
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8439036367409938e-05,
+      "loss": 0.3187,
+      "step": 1682
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8436826249059657e-05,
+      "loss": 0.2994,
+      "step": 1683
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.843461465320192e-05,
+      "loss": 0.2707,
+      "step": 1684
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.843240158007991e-05,
+      "loss": 0.2383,
+      "step": 1685
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.843018702993698e-05,
+      "loss": 0.2935,
+      "step": 1686
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.842797100301663e-05,
+      "loss": 0.2443,
+      "step": 1687
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8425753499562542e-05,
+      "loss": 0.3496,
+      "step": 1688
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.842353451981855e-05,
+      "loss": 0.3208,
+      "step": 1689
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.842131406402865e-05,
+      "loss": 0.2996,
+      "step": 1690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8419092132437008e-05,
+      "loss": 0.2686,
+      "step": 1691
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.841686872528794e-05,
+      "loss": 0.2169,
+      "step": 1692
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8414643842825936e-05,
+      "loss": 0.2906,
+      "step": 1693
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8412417485295637e-05,
+      "loss": 0.3096,
+      "step": 1694
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.841018965294186e-05,
+      "loss": 0.2762,
+      "step": 1695
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.840796034600957e-05,
+      "loss": 0.3422,
+      "step": 1696
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8405729564743904e-05,
+      "loss": 0.2834,
+      "step": 1697
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.840349730939016e-05,
+      "loss": 0.3046,
+      "step": 1698
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8401263580193785e-05,
+      "loss": 0.3468,
+      "step": 1699
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8399028377400412e-05,
+      "loss": 0.2527,
+      "step": 1700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8396791701255817e-05,
+      "loss": 0.2589,
+      "step": 1701
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8394553552005943e-05,
+      "loss": 0.3914,
+      "step": 1702
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8392313929896894e-05,
+      "loss": 0.316,
+      "step": 1703
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8390072835174948e-05,
+      "loss": 0.3022,
+      "step": 1704
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8387830268086522e-05,
+      "loss": 0.3458,
+      "step": 1705
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.838558622887822e-05,
+      "loss": 0.3519,
+      "step": 1706
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8383340717796783e-05,
+      "loss": 0.2922,
+      "step": 1707
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8381093735089137e-05,
+      "loss": 0.2737,
+      "step": 1708
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.837884528100235e-05,
+      "loss": 0.3034,
+      "step": 1709
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8376595355783665e-05,
+      "loss": 0.2694,
+      "step": 1710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.837434395968049e-05,
+      "loss": 0.329,
+      "step": 1711
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8372091092940376e-05,
+      "loss": 0.2793,
+      "step": 1712
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8369836755811056e-05,
+      "loss": 0.3417,
+      "step": 1713
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8367580948540408e-05,
+      "loss": 0.2224,
+      "step": 1714
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.836532367137649e-05,
+      "loss": 0.2603,
+      "step": 1715
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8363064924567498e-05,
+      "loss": 0.3389,
+      "step": 1716
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8360804708361815e-05,
+      "loss": 0.281,
+      "step": 1717
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.835854302300796e-05,
+      "loss": 0.3472,
+      "step": 1718
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8356279868754644e-05,
+      "loss": 0.2751,
+      "step": 1719
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.835401524585071e-05,
+      "loss": 0.3584,
+      "step": 1720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8351749154545175e-05,
+      "loss": 0.348,
+      "step": 1721
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8349481595087227e-05,
+      "loss": 0.3162,
+      "step": 1722
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8347212567726196e-05,
+      "loss": 0.2439,
+      "step": 1723
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8344942072711587e-05,
+      "loss": 0.3431,
+      "step": 1724
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8342670110293065e-05,
+      "loss": 0.2732,
+      "step": 1725
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8340396680720447e-05,
+      "loss": 0.3176,
+      "step": 1726
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8338121784243726e-05,
+      "loss": 0.3304,
+      "step": 1727
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.833584542111305e-05,
+      "loss": 0.3143,
+      "step": 1728
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8333567591578712e-05,
+      "loss": 0.349,
+      "step": 1729
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.83312882958912e-05,
+      "loss": 0.3124,
+      "step": 1730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8329007534301128e-05,
+      "loss": 0.3744,
+      "step": 1731
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.83267253070593e-05,
+      "loss": 0.2245,
+      "step": 1732
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.832444161441666e-05,
+      "loss": 0.3778,
+      "step": 1733
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8322156456624333e-05,
+      "loss": 0.3019,
+      "step": 1734
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.831986983393358e-05,
+      "loss": 0.3671,
+      "step": 1735
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.831758174659585e-05,
+      "loss": 0.3658,
+      "step": 1736
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.831529219486273e-05,
+      "loss": 0.2413,
+      "step": 1737
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.831300117898598e-05,
+      "loss": 0.2051,
+      "step": 1738
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8310708699217524e-05,
+      "loss": 0.3658,
+      "step": 1739
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8308414755809437e-05,
+      "loss": 0.3339,
+      "step": 1740
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8306119349013965e-05,
+      "loss": 0.3314,
+      "step": 1741
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8303822479083507e-05,
+      "loss": 0.3271,
+      "step": 1742
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8301524146270622e-05,
+      "loss": 0.3595,
+      "step": 1743
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.829922435082804e-05,
+      "loss": 0.2744,
+      "step": 1744
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8296923093008642e-05,
+      "loss": 0.3442,
+      "step": 1745
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8294620373065473e-05,
+      "loss": 0.323,
+      "step": 1746
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8292316191251743e-05,
+      "loss": 0.3285,
+      "step": 1747
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8290010547820814e-05,
+      "loss": 0.2959,
+      "step": 1748
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.828770344302622e-05,
+      "loss": 0.2323,
+      "step": 1749
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.828539487712164e-05,
+      "loss": 0.3417,
+      "step": 1750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8283084850360926e-05,
+      "loss": 0.3826,
+      "step": 1751
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.82807733629981e-05,
+      "loss": 0.2363,
+      "step": 1752
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.827846041528731e-05,
+      "loss": 0.296,
+      "step": 1753
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.82761460074829e-05,
+      "loss": 0.3378,
+      "step": 1754
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8273830139839362e-05,
+      "loss": 0.3138,
+      "step": 1755
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8271512812611346e-05,
+      "loss": 0.215,
+      "step": 1756
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8269194026053663e-05,
+      "loss": 0.4196,
+      "step": 1757
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.826687378042128e-05,
+      "loss": 0.3033,
+      "step": 1758
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8264552075969344e-05,
+      "loss": 0.2405,
+      "step": 1759
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.826222891295314e-05,
+      "loss": 0.367,
+      "step": 1760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.825990429162812e-05,
+      "loss": 0.2496,
+      "step": 1761
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.82575782122499e-05,
+      "loss": 0.2954,
+      "step": 1762
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8255250675074258e-05,
+      "loss": 0.3708,
+      "step": 1763
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.825292168035713e-05,
+      "loss": 0.3481,
+      "step": 1764
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.82505912283546e-05,
+      "loss": 0.4158,
+      "step": 1765
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.824825931932294e-05,
+      "loss": 0.2919,
+      "step": 1766
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8245925953518553e-05,
+      "loss": 0.2747,
+      "step": 1767
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8243591131198018e-05,
+      "loss": 0.2299,
+      "step": 1768
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8241254852618073e-05,
+      "loss": 0.3319,
+      "step": 1769
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8238917118035618e-05,
+      "loss": 0.3667,
+      "step": 1770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.82365779277077e-05,
+      "loss": 0.2761,
+      "step": 1771
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8234237281891538e-05,
+      "loss": 0.3175,
+      "step": 1772
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8231895180844512e-05,
+      "loss": 0.2925,
+      "step": 1773
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8229551624824154e-05,
+      "loss": 0.308,
+      "step": 1774
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.822720661408817e-05,
+      "loss": 0.3563,
+      "step": 1775
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8224860148894405e-05,
+      "loss": 0.2847,
+      "step": 1776
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.822251222950088e-05,
+      "loss": 0.2318,
+      "step": 1777
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.822016285616577e-05,
+      "loss": 0.3098,
+      "step": 1778
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8217812029147412e-05,
+      "loss": 0.3159,
+      "step": 1779
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8215459748704303e-05,
+      "loss": 0.2562,
+      "step": 1780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.82131060150951e-05,
+      "loss": 0.3577,
+      "step": 1781
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8210750828578607e-05,
+      "loss": 0.3582,
+      "step": 1782
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8208394189413815e-05,
+      "loss": 0.2686,
+      "step": 1783
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8206036097859853e-05,
+      "loss": 0.2402,
+      "step": 1784
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8203676554176015e-05,
+      "loss": 0.2573,
+      "step": 1785
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.820131555862175e-05,
+      "loss": 0.3488,
+      "step": 1786
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.819895311145668e-05,
+      "loss": 0.2418,
+      "step": 1787
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8196589212940577e-05,
+      "loss": 0.2489,
+      "step": 1788
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8194223863333373e-05,
+      "loss": 0.259,
+      "step": 1789
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.819185706289516e-05,
+      "loss": 0.3042,
+      "step": 1790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8189488811886195e-05,
+      "loss": 0.3188,
+      "step": 1791
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8187119110566882e-05,
+      "loss": 0.3448,
+      "step": 1792
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.81847479591978e-05,
+      "loss": 0.2378,
+      "step": 1793
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8182375358039673e-05,
+      "loss": 0.3661,
+      "step": 1794
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8180001307353393e-05,
+      "loss": 0.2585,
+      "step": 1795
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8177625807400012e-05,
+      "loss": 0.2609,
+      "step": 1796
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.817524885844074e-05,
+      "loss": 0.2892,
+      "step": 1797
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.817287046073694e-05,
+      "loss": 0.2145,
+      "step": 1798
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.817049061455015e-05,
+      "loss": 0.3367,
+      "step": 1799
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8168109320142042e-05,
+      "loss": 0.2993,
+      "step": 1800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8165726577774473e-05,
+      "loss": 0.2722,
+      "step": 1801
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.816334238770944e-05,
+      "loss": 0.3696,
+      "step": 1802
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.816095675020912e-05,
+      "loss": 0.2715,
+      "step": 1803
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8158569665535823e-05,
+      "loss": 0.2762,
+      "step": 1804
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.815618113395204e-05,
+      "loss": 0.2794,
+      "step": 1805
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.815379115572041e-05,
+      "loss": 0.2303,
+      "step": 1806
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8151399731103736e-05,
+      "loss": 0.3082,
+      "step": 1807
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8149006860364976e-05,
+      "loss": 0.298,
+      "step": 1808
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8146612543767246e-05,
+      "loss": 0.2556,
+      "step": 1809
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8144216781573832e-05,
+      "loss": 0.266,
+      "step": 1810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8141819574048168e-05,
+      "loss": 0.3535,
+      "step": 1811
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8139420921453843e-05,
+      "loss": 0.2955,
+      "step": 1812
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.813702082405462e-05,
+      "loss": 0.293,
+      "step": 1813
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.813461928211441e-05,
+      "loss": 0.2898,
+      "step": 1814
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8132216295897288e-05,
+      "loss": 0.3788,
+      "step": 1815
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8129811865667482e-05,
+      "loss": 0.2131,
+      "step": 1816
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8127405991689378e-05,
+      "loss": 0.2902,
+      "step": 1817
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.812499867422753e-05,
+      "loss": 0.3053,
+      "step": 1818
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8122589913546647e-05,
+      "loss": 0.3497,
+      "step": 1819
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8120179709911593e-05,
+      "loss": 0.2601,
+      "step": 1820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8117768063587396e-05,
+      "loss": 0.3539,
+      "step": 1821
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.811535497483923e-05,
+      "loss": 0.3793,
+      "step": 1822
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.811294044393245e-05,
+      "loss": 0.2981,
+      "step": 1823
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8110524471132553e-05,
+      "loss": 0.3315,
+      "step": 1824
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.810810705670519e-05,
+      "loss": 0.3026,
+      "step": 1825
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8105688200916186e-05,
+      "loss": 0.2955,
+      "step": 1826
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.810326790403152e-05,
+      "loss": 0.2684,
+      "step": 1827
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.810084616631732e-05,
+      "loss": 0.3349,
+      "step": 1828
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8098422988039886e-05,
+      "loss": 0.3009,
+      "step": 1829
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8095998369465656e-05,
+      "loss": 0.1954,
+      "step": 1830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.809357231086126e-05,
+      "loss": 0.2928,
+      "step": 1831
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.809114481249345e-05,
+      "loss": 0.3253,
+      "step": 1832
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.808871587462916e-05,
+      "loss": 0.2817,
+      "step": 1833
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.808628549753548e-05,
+      "loss": 0.2406,
+      "step": 1834
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8083853681479638e-05,
+      "loss": 0.3042,
+      "step": 1835
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.808142042672905e-05,
+      "loss": 0.2361,
+      "step": 1836
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8078985733551263e-05,
+      "loss": 0.2763,
+      "step": 1837
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.807654960221401e-05,
+      "loss": 0.265,
+      "step": 1838
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8074112032985153e-05,
+      "loss": 0.2621,
+      "step": 1839
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8071673026132732e-05,
+      "loss": 0.3242,
+      "step": 1840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.806923258192494e-05,
+      "loss": 0.3816,
+      "step": 1841
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.806679070063013e-05,
+      "loss": 0.2435,
+      "step": 1842
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.80643473825168e-05,
+      "loss": 0.2988,
+      "step": 1843
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8061902627853623e-05,
+      "loss": 0.2341,
+      "step": 1844
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8059456436909424e-05,
+      "loss": 0.3172,
+      "step": 1845
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8057008809953183e-05,
+      "loss": 0.2476,
+      "step": 1846
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8054559747254043e-05,
+      "loss": 0.3001,
+      "step": 1847
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8052109249081297e-05,
+      "loss": 0.3546,
+      "step": 1848
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.80496573157044e-05,
+      "loss": 0.2458,
+      "step": 1849
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.804720394739297e-05,
+      "loss": 0.1948,
+      "step": 1850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.804474914441678e-05,
+      "loss": 0.2482,
+      "step": 1851
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8042292907045756e-05,
+      "loss": 0.3354,
+      "step": 1852
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.803983523554998e-05,
+      "loss": 0.296,
+      "step": 1853
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8037376130199702e-05,
+      "loss": 0.2927,
+      "step": 1854
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8034915591265327e-05,
+      "loss": 0.2638,
+      "step": 1855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8032453619017405e-05,
+      "loss": 0.2527,
+      "step": 1856
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8029990213726667e-05,
+      "loss": 0.2548,
+      "step": 1857
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8027525375663978e-05,
+      "loss": 0.2375,
+      "step": 1858
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8025059105100368e-05,
+      "loss": 0.2412,
+      "step": 1859
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.802259140230704e-05,
+      "loss": 0.2867,
+      "step": 1860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8020122267555326e-05,
+      "loss": 0.2877,
+      "step": 1861
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8017651701116737e-05,
+      "loss": 0.4032,
+      "step": 1862
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8015179703262943e-05,
+      "loss": 0.2484,
+      "step": 1863
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8012706274265755e-05,
+      "loss": 0.2664,
+      "step": 1864
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.801023141439715e-05,
+      "loss": 0.3154,
+      "step": 1865
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.800775512392927e-05,
+      "loss": 0.3864,
+      "step": 1866
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8005277403134396e-05,
+      "loss": 0.3644,
+      "step": 1867
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.800279825228499e-05,
+      "loss": 0.2325,
+      "step": 1868
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8000317671653647e-05,
+      "loss": 0.2273,
+      "step": 1869
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7997835661513135e-05,
+      "loss": 0.2631,
+      "step": 1870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7995352222136378e-05,
+      "loss": 0.3169,
+      "step": 1871
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7992867353796447e-05,
+      "loss": 0.3475,
+      "step": 1872
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7990381056766583e-05,
+      "loss": 0.326,
+      "step": 1873
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7987893331320174e-05,
+      "loss": 0.2665,
+      "step": 1874
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7985404177730767e-05,
+      "loss": 0.3264,
+      "step": 1875
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7982913596272077e-05,
+      "loss": 0.2638,
+      "step": 1876
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7980421587217962e-05,
+      "loss": 0.2928,
+      "step": 1877
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.797792815084244e-05,
+      "loss": 0.2205,
+      "step": 1878
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7975433287419696e-05,
+      "loss": 0.241,
+      "step": 1879
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7972936997224052e-05,
+      "loss": 0.3572,
+      "step": 1880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.797043928053001e-05,
+      "loss": 0.1798,
+      "step": 1881
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.796794013761221e-05,
+      "loss": 0.2791,
+      "step": 1882
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.796543956874546e-05,
+      "loss": 0.3578,
+      "step": 1883
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7962937574204727e-05,
+      "loss": 0.1623,
+      "step": 1884
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7960434154265124e-05,
+      "loss": 0.2706,
+      "step": 1885
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.795792930920192e-05,
+      "loss": 0.2638,
+      "step": 1886
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7955423039290558e-05,
+      "loss": 0.2653,
+      "step": 1887
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7952915344806617e-05,
+      "loss": 0.2942,
+      "step": 1888
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7950406226025845e-05,
+      "loss": 0.2794,
+      "step": 1889
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7947895683224145e-05,
+      "loss": 0.3171,
+      "step": 1890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7945383716677577e-05,
+      "loss": 0.2405,
+      "step": 1891
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7942870326662353e-05,
+      "loss": 0.2686,
+      "step": 1892
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7940355513454843e-05,
+      "loss": 0.3083,
+      "step": 1893
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7937839277331572e-05,
+      "loss": 0.2841,
+      "step": 1894
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7935321618569236e-05,
+      "loss": 0.2605,
+      "step": 1895
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7932802537444665e-05,
+      "loss": 0.2199,
+      "step": 1896
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7930282034234856e-05,
+      "loss": 0.2661,
+      "step": 1897
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.792776010921697e-05,
+      "loss": 0.3279,
+      "step": 1898
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.792523676266831e-05,
+      "loss": 0.2885,
+      "step": 1899
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.792271199486635e-05,
+      "loss": 0.2573,
+      "step": 1900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.79201858060887e-05,
+      "loss": 0.3488,
+      "step": 1901
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7917658196613143e-05,
+      "loss": 0.2586,
+      "step": 1902
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.791512916671762e-05,
+      "loss": 0.3062,
+      "step": 1903
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7912598716680217e-05,
+      "loss": 0.2196,
+      "step": 1904
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7910066846779186e-05,
+      "loss": 0.2979,
+      "step": 1905
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7907533557292927e-05,
+      "loss": 0.3048,
+      "step": 1906
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7904998848499994e-05,
+      "loss": 0.3022,
+      "step": 1907
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7902462720679113e-05,
+      "loss": 0.3143,
+      "step": 1908
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7899925174109142e-05,
+      "loss": 0.281,
+      "step": 1909
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.789738620906912e-05,
+      "loss": 0.4775,
+      "step": 1910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.789484582583823e-05,
+      "loss": 0.3088,
+      "step": 1911
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7892304024695806e-05,
+      "loss": 0.2899,
+      "step": 1912
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.788976080592134e-05,
+      "loss": 0.2374,
+      "step": 1913
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7887216169794497e-05,
+      "loss": 0.2982,
+      "step": 1914
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7884670116595073e-05,
+      "loss": 0.3218,
+      "step": 1915
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7882122646603036e-05,
+      "loss": 0.2679,
+      "step": 1916
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.78795737600985e-05,
+      "loss": 0.1576,
+      "step": 1917
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7877023457361746e-05,
+      "loss": 0.2974,
+      "step": 1918
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7874471738673194e-05,
+      "loss": 0.2683,
+      "step": 1919
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7871918604313436e-05,
+      "loss": 0.2378,
+      "step": 1920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.786936405456322e-05,
+      "loss": 0.2763,
+      "step": 1921
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7866808089703433e-05,
+      "loss": 0.2936,
+      "step": 1922
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7864250710015133e-05,
+      "loss": 0.3394,
+      "step": 1923
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.786169191577952e-05,
+      "loss": 0.2166,
+      "step": 1924
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7859131707277966e-05,
+      "loss": 0.2326,
+      "step": 1925
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7856570084791998e-05,
+      "loss": 0.3431,
+      "step": 1926
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7854007048603273e-05,
+      "loss": 0.2516,
+      "step": 1927
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7851442598993634e-05,
+      "loss": 0.2197,
+      "step": 1928
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.784887673624506e-05,
+      "loss": 0.2873,
+      "step": 1929
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7846309460639698e-05,
+      "loss": 0.2241,
+      "step": 1930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7843740772459836e-05,
+      "loss": 0.3062,
+      "step": 1931
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7841170671987942e-05,
+      "loss": 0.3551,
+      "step": 1932
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7838599159506603e-05,
+      "loss": 0.2486,
+      "step": 1933
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7836026235298594e-05,
+      "loss": 0.3055,
+      "step": 1934
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.783345189964683e-05,
+      "loss": 0.2543,
+      "step": 1935
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7830876152834382e-05,
+      "loss": 0.2688,
+      "step": 1936
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7828298995144478e-05,
+      "loss": 0.2604,
+      "step": 1937
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7825720426860504e-05,
+      "loss": 0.2783,
+      "step": 1938
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7823140448265998e-05,
+      "loss": 0.2936,
+      "step": 1939
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.782055905964465e-05,
+      "loss": 0.2872,
+      "step": 1940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.781797626128031e-05,
+      "loss": 0.27,
+      "step": 1941
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.781539205345698e-05,
+      "loss": 0.2979,
+      "step": 1942
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7812806436458824e-05,
+      "loss": 0.3148,
+      "step": 1943
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7810219410570146e-05,
+      "loss": 0.2615,
+      "step": 1944
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7807630976075425e-05,
+      "loss": 0.2563,
+      "step": 1945
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7805041133259273e-05,
+      "loss": 0.2761,
+      "step": 1946
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7802449882406477e-05,
+      "loss": 0.309,
+      "step": 1947
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7799857223801968e-05,
+      "loss": 0.2494,
+      "step": 1948
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.779726315773083e-05,
+      "loss": 0.3069,
+      "step": 1949
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7794667684478305e-05,
+      "loss": 0.1654,
+      "step": 1950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7792070804329797e-05,
+      "loss": 0.2681,
+      "step": 1951
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.778947251757085e-05,
+      "loss": 0.313,
+      "step": 1952
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7786872824487173e-05,
+      "loss": 0.3381,
+      "step": 1953
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7784271725364628e-05,
+      "loss": 0.2678,
+      "step": 1954
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.778166922048923e-05,
+      "loss": 0.2584,
+      "step": 1955
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7779065310147152e-05,
+      "loss": 0.356,
+      "step": 1956
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7776459994624717e-05,
+      "loss": 0.2508,
+      "step": 1957
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.77738532742084e-05,
+      "loss": 0.4049,
+      "step": 1958
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.777124514918484e-05,
+      "loss": 0.3314,
+      "step": 1959
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7768635619840824e-05,
+      "loss": 0.3123,
+      "step": 1960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7766024686463294e-05,
+      "loss": 0.3157,
+      "step": 1961
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.776341234933935e-05,
+      "loss": 0.3564,
+      "step": 1962
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7760798608756238e-05,
+      "loss": 0.3212,
+      "step": 1963
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.775818346500137e-05,
+      "loss": 0.3096,
+      "step": 1964
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7755566918362303e-05,
+      "loss": 0.2786,
+      "step": 1965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7752948969126747e-05,
+      "loss": 0.2745,
+      "step": 1966
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.775032961758258e-05,
+      "loss": 0.2347,
+      "step": 1967
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7747708864017815e-05,
+      "loss": 0.2604,
+      "step": 1968
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7745086708720637e-05,
+      "loss": 0.2436,
+      "step": 1969
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.774246315197937e-05,
+      "loss": 0.2151,
+      "step": 1970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7739838194082505e-05,
+      "loss": 0.3125,
+      "step": 1971
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.773721183531868e-05,
+      "loss": 0.2652,
+      "step": 1972
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7734584075976684e-05,
+      "loss": 0.2328,
+      "step": 1973
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.773195491634547e-05,
+      "loss": 0.2806,
+      "step": 1974
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.772932435671413e-05,
+      "loss": 0.2505,
+      "step": 1975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7726692397371936e-05,
+      "loss": 0.3201,
+      "step": 1976
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.772405903860828e-05,
+      "loss": 0.2963,
+      "step": 1977
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7721424280712734e-05,
+      "loss": 0.3354,
+      "step": 1978
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7718788123975015e-05,
+      "loss": 0.3882,
+      "step": 1979
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7716150568684985e-05,
+      "loss": 0.2112,
+      "step": 1980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.771351161513268e-05,
+      "loss": 0.2646,
+      "step": 1981
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7710871263608268e-05,
+      "loss": 0.3829,
+      "step": 1982
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7708229514402092e-05,
+      "loss": 0.2516,
+      "step": 1983
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7705586367804626e-05,
+      "loss": 0.3295,
+      "step": 1984
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.770294182410651e-05,
+      "loss": 0.2771,
+      "step": 1985
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.770029588359855e-05,
+      "loss": 0.3815,
+      "step": 1986
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.769764854657168e-05,
+      "loss": 0.3267,
+      "step": 1987
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7694999813317006e-05,
+      "loss": 0.3334,
+      "step": 1988
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7692349684125777e-05,
+      "loss": 0.403,
+      "step": 1989
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7689698159289402e-05,
+      "loss": 0.2145,
+      "step": 1990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.768704523909944e-05,
+      "loss": 0.2543,
+      "step": 1991
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.768439092384761e-05,
+      "loss": 0.2393,
+      "step": 1992
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7681735213825775e-05,
+      "loss": 0.2842,
+      "step": 1993
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7679078109325957e-05,
+      "loss": 0.1319,
+      "step": 1994
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7676419610640333e-05,
+      "loss": 0.3239,
+      "step": 1995
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.767375971806122e-05,
+      "loss": 0.2676,
+      "step": 1996
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.767109843188111e-05,
+      "loss": 0.24,
+      "step": 1997
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7668435752392635e-05,
+      "loss": 0.2158,
+      "step": 1998
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7665771679888575e-05,
+      "loss": 0.2771,
+      "step": 1999
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.766310621466188e-05,
+      "loss": 0.2806,
+      "step": 2000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7660439357005628e-05,
+      "loss": 0.3099,
+      "step": 2001
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.765777110721309e-05,
+      "loss": 0.2946,
+      "step": 2002
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7655101465577638e-05,
+      "loss": 0.3414,
+      "step": 2003
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7652430432392845e-05,
+      "loss": 0.3405,
+      "step": 2004
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.764975800795241e-05,
+      "loss": 0.3656,
+      "step": 2005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7647084192550187e-05,
+      "loss": 0.2695,
+      "step": 2006
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7644408986480196e-05,
+      "loss": 0.3323,
+      "step": 2007
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7641732390036596e-05,
+      "loss": 0.2594,
+      "step": 2008
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.76390544035137e-05,
+      "loss": 0.3574,
+      "step": 2009
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.763637502720599e-05,
+      "loss": 0.2235,
+      "step": 2010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7633694261408086e-05,
+      "loss": 0.28,
+      "step": 2011
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.763101210641475e-05,
+      "loss": 0.2881,
+      "step": 2012
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.762832856252093e-05,
+      "loss": 0.3112,
+      "step": 2013
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7625643630021696e-05,
+      "loss": 0.2936,
+      "step": 2014
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7622957309212283e-05,
+      "loss": 0.2512,
+      "step": 2015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7620269600388082e-05,
+      "loss": 0.3118,
+      "step": 2016
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7617580503844623e-05,
+      "loss": 0.2147,
+      "step": 2017
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7614890019877608e-05,
+      "loss": 0.2096,
+      "step": 2018
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.761219814878288e-05,
+      "loss": 0.3368,
+      "step": 2019
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7609504890856434e-05,
+      "loss": 0.3271,
+      "step": 2020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7606810246394416e-05,
+      "loss": 0.2811,
+      "step": 2021
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.760411421569313e-05,
+      "loss": 0.1878,
+      "step": 2022
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7601416799049032e-05,
+      "loss": 0.2709,
+      "step": 2023
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.759871799675873e-05,
+      "loss": 0.2857,
+      "step": 2024
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7596017809118978e-05,
+      "loss": 0.294,
+      "step": 2025
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7593316236426695e-05,
+      "loss": 0.3795,
+      "step": 2026
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.759061327897894e-05,
+      "loss": 0.3932,
+      "step": 2027
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7587908937072925e-05,
+      "loss": 0.202,
+      "step": 2028
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7585203211006024e-05,
+      "loss": 0.2653,
+      "step": 2029
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.758249610107576e-05,
+      "loss": 0.3201,
+      "step": 2030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7579787607579797e-05,
+      "loss": 0.2195,
+      "step": 2031
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7577077730815966e-05,
+      "loss": 0.3457,
+      "step": 2032
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.757436647108224e-05,
+      "loss": 0.2394,
+      "step": 2033
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.757165382867675e-05,
+      "loss": 0.3127,
+      "step": 2034
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.756893980389778e-05,
+      "loss": 0.2324,
+      "step": 2035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7566224397043763e-05,
+      "loss": 0.2076,
+      "step": 2036
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7563507608413273e-05,
+      "loss": 0.2661,
+      "step": 2037
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.756078943830506e-05,
+      "loss": 0.1723,
+      "step": 2038
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7558069887018006e-05,
+      "loss": 0.207,
+      "step": 2039
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7555348954851153e-05,
+      "loss": 0.2504,
+      "step": 2040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.75526266421037e-05,
+      "loss": 0.386,
+      "step": 2041
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7549902949074973e-05,
+      "loss": 0.2994,
+      "step": 2042
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.754717787606449e-05,
+      "loss": 0.3754,
+      "step": 2043
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7544451423371885e-05,
+      "loss": 0.2454,
+      "step": 2044
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7541723591296964e-05,
+      "loss": 0.2694,
+      "step": 2045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.753899438013967e-05,
+      "loss": 0.1924,
+      "step": 2046
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7536263790200114e-05,
+      "loss": 0.3146,
+      "step": 2047
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7533531821778553e-05,
+      "loss": 0.2693,
+      "step": 2048
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7530798475175385e-05,
+      "loss": 0.3516,
+      "step": 2049
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7528063750691166e-05,
+      "loss": 0.2805,
+      "step": 2050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7525327648626616e-05,
+      "loss": 0.2791,
+      "step": 2051
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7522590169282586e-05,
+      "loss": 0.2267,
+      "step": 2052
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7519851312960093e-05,
+      "loss": 0.2628,
+      "step": 2053
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7517111079960298e-05,
+      "loss": 0.2616,
+      "step": 2054
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7514369470584516e-05,
+      "loss": 0.2524,
+      "step": 2055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7511626485134214e-05,
+      "loss": 0.3033,
+      "step": 2056
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.750888212391101e-05,
+      "loss": 0.3396,
+      "step": 2057
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.750613638721667e-05,
+      "loss": 0.2321,
+      "step": 2058
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.750338927535312e-05,
+      "loss": 0.3053,
+      "step": 2059
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7500640788622424e-05,
+      "loss": 0.2905,
+      "step": 2060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7497890927326808e-05,
+      "loss": 0.3148,
+      "step": 2061
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7495139691768644e-05,
+      "loss": 0.36,
+      "step": 2062
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7492387082250456e-05,
+      "loss": 0.2419,
+      "step": 2063
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7489633099074922e-05,
+      "loss": 0.2787,
+      "step": 2064
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.748687774254487e-05,
+      "loss": 0.3255,
+      "step": 2065
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7484121012963278e-05,
+      "loss": 0.2677,
+      "step": 2066
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.748136291063327e-05,
+      "loss": 0.2496,
+      "step": 2067
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7478603435858126e-05,
+      "loss": 0.3491,
+      "step": 2068
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7475842588941283e-05,
+      "loss": 0.2071,
+      "step": 2069
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.747308037018632e-05,
+      "loss": 0.4125,
+      "step": 2070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7470316779896967e-05,
+      "loss": 0.2733,
+      "step": 2071
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7467551818377105e-05,
+      "loss": 0.3258,
+      "step": 2072
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.746478548593077e-05,
+      "loss": 0.3326,
+      "step": 2073
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.746201778286215e-05,
+      "loss": 0.258,
+      "step": 2074
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7459248709475577e-05,
+      "loss": 0.2968,
+      "step": 2075
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.745647826607554e-05,
+      "loss": 0.2728,
+      "step": 2076
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7453706452966675e-05,
+      "loss": 0.252,
+      "step": 2077
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7450933270453766e-05,
+      "loss": 0.2595,
+      "step": 2078
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7448158718841752e-05,
+      "loss": 0.251,
+      "step": 2079
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7445382798435727e-05,
+      "loss": 0.1746,
+      "step": 2080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.744260550954092e-05,
+      "loss": 0.2917,
+      "step": 2081
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.743982685246273e-05,
+      "loss": 0.2396,
+      "step": 2082
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7437046827506687e-05,
+      "loss": 0.3002,
+      "step": 2083
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7434265434978493e-05,
+      "loss": 0.2208,
+      "step": 2084
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7431482675183983e-05,
+      "loss": 0.3135,
+      "step": 2085
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7428698548429146e-05,
+      "loss": 0.2489,
+      "step": 2086
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.742591305502013e-05,
+      "loss": 0.3058,
+      "step": 2087
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7423126195263215e-05,
+      "loss": 0.3139,
+      "step": 2088
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.742033796946485e-05,
+      "loss": 0.2477,
+      "step": 2089
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7417548377931628e-05,
+      "loss": 0.2684,
+      "step": 2090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7414757420970296e-05,
+      "loss": 0.3043,
+      "step": 2091
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.741196509888773e-05,
+      "loss": 0.2393,
+      "step": 2092
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7409171411990994e-05,
+      "loss": 0.2572,
+      "step": 2093
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7406376360587262e-05,
+      "loss": 0.2565,
+      "step": 2094
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.740357994498389e-05,
+      "loss": 0.2697,
+      "step": 2095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.740078216548836e-05,
+      "loss": 0.291,
+      "step": 2096
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7397983022408322e-05,
+      "loss": 0.333,
+      "step": 2097
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7395182516051566e-05,
+      "loss": 0.226,
+      "step": 2098
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7392380646726034e-05,
+      "loss": 0.2363,
+      "step": 2099
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7389577414739816e-05,
+      "loss": 0.2463,
+      "step": 2100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7386772820401158e-05,
+      "loss": 0.3829,
+      "step": 2101
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.738396686401845e-05,
+      "loss": 0.275,
+      "step": 2102
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7381159545900238e-05,
+      "loss": 0.2672,
+      "step": 2103
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7378350866355207e-05,
+      "loss": 0.2327,
+      "step": 2104
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7375540825692205e-05,
+      "loss": 0.2163,
+      "step": 2105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7372729424220215e-05,
+      "loss": 0.2213,
+      "step": 2106
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7369916662248384e-05,
+      "loss": 0.3072,
+      "step": 2107
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7367102540085998e-05,
+      "loss": 0.2494,
+      "step": 2108
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.73642870580425e-05,
+      "loss": 0.182,
+      "step": 2109
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7361470216427472e-05,
+      "loss": 0.2234,
+      "step": 2110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7358652015550662e-05,
+      "loss": 0.3246,
+      "step": 2111
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7355832455721955e-05,
+      "loss": 0.228,
+      "step": 2112
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7353011537251386e-05,
+      "loss": 0.1937,
+      "step": 2113
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7350189260449137e-05,
+      "loss": 0.2183,
+      "step": 2114
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.734736562562556e-05,
+      "loss": 0.2272,
+      "step": 2115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.734454063309113e-05,
+      "loss": 0.3419,
+      "step": 2116
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.734171428315648e-05,
+      "loss": 0.2443,
+      "step": 2117
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7338886576132402e-05,
+      "loss": 0.3555,
+      "step": 2118
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7336057512329823e-05,
+      "loss": 0.2384,
+      "step": 2119
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7333227092059825e-05,
+      "loss": 0.256,
+      "step": 2120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7330395315633646e-05,
+      "loss": 0.295,
+      "step": 2121
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7327562183362662e-05,
+      "loss": 0.2772,
+      "step": 2122
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7324727695558404e-05,
+      "loss": 0.2903,
+      "step": 2123
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.732189185253255e-05,
+      "loss": 0.2354,
+      "step": 2124
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.731905465459693e-05,
+      "loss": 0.241,
+      "step": 2125
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.731621610206352e-05,
+      "loss": 0.2155,
+      "step": 2126
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7313376195244448e-05,
+      "loss": 0.3,
+      "step": 2127
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7310534934451984e-05,
+      "loss": 0.3774,
+      "step": 2128
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7307692319998562e-05,
+      "loss": 0.2991,
+      "step": 2129
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7304848352196745e-05,
+      "loss": 0.3439,
+      "step": 2130
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.730200303135925e-05,
+      "loss": 0.2692,
+      "step": 2131
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.729915635779896e-05,
+      "loss": 0.2328,
+      "step": 2132
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7296308331828888e-05,
+      "loss": 0.2817,
+      "step": 2133
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.72934589537622e-05,
+      "loss": 0.272,
+      "step": 2134
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7290608223912215e-05,
+      "loss": 0.2767,
+      "step": 2135
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7287756142592396e-05,
+      "loss": 0.2504,
+      "step": 2136
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.728490271011636e-05,
+      "loss": 0.2532,
+      "step": 2137
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7282047926797863e-05,
+      "loss": 0.3663,
+      "step": 2138
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7279191792950825e-05,
+      "loss": 0.2814,
+      "step": 2139
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.72763343088893e-05,
+      "loss": 0.3466,
+      "step": 2140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.727347547492749e-05,
+      "loss": 0.2306,
+      "step": 2141
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.727061529137976e-05,
+      "loss": 0.2392,
+      "step": 2142
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7267753758560608e-05,
+      "loss": 0.2833,
+      "step": 2143
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7264890876784698e-05,
+      "loss": 0.2971,
+      "step": 2144
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7262026646366816e-05,
+      "loss": 0.3009,
+      "step": 2145
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.725916106762192e-05,
+      "loss": 0.1912,
+      "step": 2146
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.725629414086511e-05,
+      "loss": 0.2172,
+      "step": 2147
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7253425866411625e-05,
+      "loss": 0.2853,
+      "step": 2148
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7250556244576863e-05,
+      "loss": 0.2001,
+      "step": 2149
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.724768527567637e-05,
+      "loss": 0.2255,
+      "step": 2150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7244812960025828e-05,
+      "loss": 0.2355,
+      "step": 2151
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.724193929794108e-05,
+      "loss": 0.3721,
+      "step": 2152
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7239064289738116e-05,
+      "loss": 0.4087,
+      "step": 2153
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7236187935733065e-05,
+      "loss": 0.3055,
+      "step": 2154
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7233310236242215e-05,
+      "loss": 0.353,
+      "step": 2155
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.723043119158199e-05,
+      "loss": 0.2731,
+      "step": 2156
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7227550802068972e-05,
+      "loss": 0.1755,
+      "step": 2157
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.722466906801989e-05,
+      "loss": 0.2399,
+      "step": 2158
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7221785989751612e-05,
+      "loss": 0.304,
+      "step": 2159
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7218901567581166e-05,
+      "loss": 0.2958,
+      "step": 2160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7216015801825717e-05,
+      "loss": 0.3073,
+      "step": 2161
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7213128692802588e-05,
+      "loss": 0.248,
+      "step": 2162
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7210240240829243e-05,
+      "loss": 0.1488,
+      "step": 2163
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.720735044622329e-05,
+      "loss": 0.2414,
+      "step": 2164
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.720445930930249e-05,
+      "loss": 0.3549,
+      "step": 2165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.720156683038476e-05,
+      "loss": 0.3376,
+      "step": 2166
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.719867300978814e-05,
+      "loss": 0.3185,
+      "step": 2167
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.719577784783085e-05,
+      "loss": 0.1971,
+      "step": 2168
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7192881344831233e-05,
+      "loss": 0.2866,
+      "step": 2169
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7189983501107784e-05,
+      "loss": 0.1891,
+      "step": 2170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7187084316979155e-05,
+      "loss": 0.3261,
+      "step": 2171
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7184183792764132e-05,
+      "loss": 0.3038,
+      "step": 2172
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7181281928781664e-05,
+      "loss": 0.2779,
+      "step": 2173
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.717837872535083e-05,
+      "loss": 0.2249,
+      "step": 2174
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.717547418279088e-05,
+      "loss": 0.2918,
+      "step": 2175
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7172568301421174e-05,
+      "loss": 0.2049,
+      "step": 2176
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7169661081561255e-05,
+      "loss": 0.1748,
+      "step": 2177
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7166752523530794e-05,
+      "loss": 0.3698,
+      "step": 2178
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.716384262764963e-05,
+      "loss": 0.2068,
+      "step": 2179
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.716093139423771e-05,
+      "loss": 0.2675,
+      "step": 2180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.715801882361517e-05,
+      "loss": 0.3823,
+      "step": 2181
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7155104916102268e-05,
+      "loss": 0.201,
+      "step": 2182
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.715218967201941e-05,
+      "loss": 0.2572,
+      "step": 2183
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7149273091687173e-05,
+      "loss": 0.2227,
+      "step": 2184
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7146355175426244e-05,
+      "loss": 0.2642,
+      "step": 2185
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7143435923557486e-05,
+      "loss": 0.2202,
+      "step": 2186
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7140515336401896e-05,
+      "loss": 0.3915,
+      "step": 2187
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7137593414280623e-05,
+      "loss": 0.2501,
+      "step": 2188
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7134670157514952e-05,
+      "loss": 0.2687,
+      "step": 2189
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7131745566426334e-05,
+      "loss": 0.1979,
+      "step": 2190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.712881964133635e-05,
+      "loss": 0.3279,
+      "step": 2191
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7125892382566732e-05,
+      "loss": 0.2492,
+      "step": 2192
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7122963790439362e-05,
+      "loss": 0.2749,
+      "step": 2193
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7120033865276266e-05,
+      "loss": 0.3691,
+      "step": 2194
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7117102607399622e-05,
+      "loss": 0.2003,
+      "step": 2195
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7114170017131746e-05,
+      "loss": 0.2282,
+      "step": 2196
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.71112360947951e-05,
+      "loss": 0.2267,
+      "step": 2197
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7108300840712305e-05,
+      "loss": 0.303,
+      "step": 2198
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7105364255206115e-05,
+      "loss": 0.2501,
+      "step": 2199
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7102426338599433e-05,
+      "loss": 0.2676,
+      "step": 2200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.709948709121532e-05,
+      "loss": 0.1532,
+      "step": 2201
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.709654651337697e-05,
+      "loss": 0.2547,
+      "step": 2202
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.709360460540772e-05,
+      "loss": 0.2839,
+      "step": 2203
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.709066136763107e-05,
+      "loss": 0.2346,
+      "step": 2204
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.708771680037066e-05,
+      "loss": 0.2768,
+      "step": 2205
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7084770903950265e-05,
+      "loss": 0.319,
+      "step": 2206
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7081823678693818e-05,
+      "loss": 0.2999,
+      "step": 2207
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7078875124925394e-05,
+      "loss": 0.2518,
+      "step": 2208
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7075925242969213e-05,
+      "loss": 0.2712,
+      "step": 2209
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7072974033149646e-05,
+      "loss": 0.2362,
+      "step": 2210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7070021495791204e-05,
+      "loss": 0.3312,
+      "step": 2211
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7067067631218554e-05,
+      "loss": 0.189,
+      "step": 2212
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.706411243975649e-05,
+      "loss": 0.2563,
+      "step": 2213
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7061155921729972e-05,
+      "loss": 0.2101,
+      "step": 2214
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.705819807746409e-05,
+      "loss": 0.2241,
+      "step": 2215
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7055238907284097e-05,
+      "loss": 0.3424,
+      "step": 2216
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7052278411515374e-05,
+      "loss": 0.3728,
+      "step": 2217
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.704931659048346e-05,
+      "loss": 0.3205,
+      "step": 2218
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7046353444514035e-05,
+      "loss": 0.212,
+      "step": 2219
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7043388973932923e-05,
+      "loss": 0.3558,
+      "step": 2220
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7040423179066095e-05,
+      "loss": 0.2573,
+      "step": 2221
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.703745606023968e-05,
+      "loss": 0.2251,
+      "step": 2222
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.703448761777992e-05,
+      "loss": 0.2594,
+      "step": 2223
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7031517852013247e-05,
+      "loss": 0.3063,
+      "step": 2224
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7028546763266193e-05,
+      "loss": 0.3394,
+      "step": 2225
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7025574351865475e-05,
+      "loss": 0.2368,
+      "step": 2226
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.702260061813793e-05,
+      "loss": 0.3057,
+      "step": 2227
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7019625562410547e-05,
+      "loss": 0.2173,
+      "step": 2228
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7016649185010467e-05,
+      "loss": 0.2629,
+      "step": 2229
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.701367148626497e-05,
+      "loss": 0.2318,
+      "step": 2230
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7010692466501478e-05,
+      "loss": 0.256,
+      "step": 2231
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7007712126047568e-05,
+      "loss": 0.1857,
+      "step": 2232
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7004730465230954e-05,
+      "loss": 0.2208,
+      "step": 2233
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7001747484379502e-05,
+      "loss": 0.2446,
+      "step": 2234
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6998763183821212e-05,
+      "loss": 0.3057,
+      "step": 2235
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.699577756388424e-05,
+      "loss": 0.3533,
+      "step": 2236
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6992790624896883e-05,
+      "loss": 0.2559,
+      "step": 2237
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.698980236718759e-05,
+      "loss": 0.2864,
+      "step": 2238
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6986812791084937e-05,
+      "loss": 0.2845,
+      "step": 2239
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6983821896917663e-05,
+      "loss": 0.2208,
+      "step": 2240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.698082968501465e-05,
+      "loss": 0.3111,
+      "step": 2241
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.697783615570491e-05,
+      "loss": 0.3243,
+      "step": 2242
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6974841309317618e-05,
+      "loss": 0.2961,
+      "step": 2243
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.697184514618208e-05,
+      "loss": 0.2992,
+      "step": 2244
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6968847666627757e-05,
+      "loss": 0.2699,
+      "step": 2245
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6965848870984254e-05,
+      "loss": 0.299,
+      "step": 2246
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6962848759581305e-05,
+      "loss": 0.2504,
+      "step": 2247
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6959847332748814e-05,
+      "loss": 0.1908,
+      "step": 2248
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6956844590816814e-05,
+      "loss": 0.2142,
+      "step": 2249
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6953840534115475e-05,
+      "loss": 0.2031,
+      "step": 2250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6950835162975135e-05,
+      "loss": 0.3436,
+      "step": 2251
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6947828477726253e-05,
+      "loss": 0.2731,
+      "step": 2252
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6944820478699445e-05,
+      "loss": 0.321,
+      "step": 2253
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.694181116622548e-05,
+      "loss": 0.322,
+      "step": 2254
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6938800540635245e-05,
+      "loss": 0.2643,
+      "step": 2255
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.69357886022598e-05,
+      "loss": 0.2841,
+      "step": 2256
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6932775351430327e-05,
+      "loss": 0.276,
+      "step": 2257
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.692976078847816e-05,
+      "loss": 0.3123,
+      "step": 2258
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.692674491373479e-05,
+      "loss": 0.1948,
+      "step": 2259
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6923727727531836e-05,
+      "loss": 0.205,
+      "step": 2260
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6920709230201067e-05,
+      "loss": 0.2767,
+      "step": 2261
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.691768942207439e-05,
+      "loss": 0.2454,
+      "step": 2262
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.691466830348387e-05,
+      "loss": 0.2733,
+      "step": 2263
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.69116458747617e-05,
+      "loss": 0.2595,
+      "step": 2264
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6908622136240237e-05,
+      "loss": 0.3376,
+      "step": 2265
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6905597088251957e-05,
+      "loss": 0.2212,
+      "step": 2266
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.69025707311295e-05,
+      "loss": 0.2477,
+      "step": 2267
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6899543065205633e-05,
+      "loss": 0.2609,
+      "step": 2268
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6896514090813296e-05,
+      "loss": 0.3043,
+      "step": 2269
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6893483808285536e-05,
+      "loss": 0.2833,
+      "step": 2270
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6890452217955565e-05,
+      "loss": 0.2832,
+      "step": 2271
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6887419320156748e-05,
+      "loss": 0.2171,
+      "step": 2272
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.688438511522256e-05,
+      "loss": 0.1811,
+      "step": 2273
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6881349603486662e-05,
+      "loss": 0.1564,
+      "step": 2274
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.687831278528282e-05,
+      "loss": 0.3328,
+      "step": 2275
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6875274660944978e-05,
+      "loss": 0.2924,
+      "step": 2276
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.687223523080719e-05,
+      "loss": 0.2258,
+      "step": 2277
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.686919449520368e-05,
+      "loss": 0.3555,
+      "step": 2278
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.68661524544688e-05,
+      "loss": 0.2839,
+      "step": 2279
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6863109108937065e-05,
+      "loss": 0.3032,
+      "step": 2280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6860064458943104e-05,
+      "loss": 0.4031,
+      "step": 2281
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6857018504821716e-05,
+      "loss": 0.2126,
+      "step": 2282
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6853971246907822e-05,
+      "loss": 0.2402,
+      "step": 2283
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.685092268553651e-05,
+      "loss": 0.3052,
+      "step": 2284
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.684787282104299e-05,
+      "loss": 0.2603,
+      "step": 2285
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6844821653762622e-05,
+      "loss": 0.2703,
+      "step": 2286
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6841769184030918e-05,
+      "loss": 0.2289,
+      "step": 2287
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6838715412183524e-05,
+      "loss": 0.2913,
+      "step": 2288
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.683566033855623e-05,
+      "loss": 0.2562,
+      "step": 2289
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.683260396348497e-05,
+      "loss": 0.2339,
+      "step": 2290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.682954628730582e-05,
+      "loss": 0.2802,
+      "step": 2291
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.682648731035501e-05,
+      "loss": 0.2408,
+      "step": 2292
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6823427032968893e-05,
+      "loss": 0.3079,
+      "step": 2293
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.682036545548398e-05,
+      "loss": 0.1832,
+      "step": 2294
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6817302578236925e-05,
+      "loss": 0.1989,
+      "step": 2295
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.681423840156451e-05,
+      "loss": 0.2094,
+      "step": 2296
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.681117292580368e-05,
+      "loss": 0.2712,
+      "step": 2297
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.680810615129151e-05,
+      "loss": 0.2266,
+      "step": 2298
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.680503807836522e-05,
+      "loss": 0.2681,
+      "step": 2299
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6801968707362172e-05,
+      "loss": 0.2,
+      "step": 2300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.679889803861988e-05,
+      "loss": 0.3076,
+      "step": 2301
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.679582607247599e-05,
+      "loss": 0.2518,
+      "step": 2302
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.679275280926829e-05,
+      "loss": 0.2372,
+      "step": 2303
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6789678249334724e-05,
+      "loss": 0.2663,
+      "step": 2304
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6786602393013358e-05,
+      "loss": 0.2369,
+      "step": 2305
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6783525240642416e-05,
+      "loss": 0.3103,
+      "step": 2306
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6780446792560266e-05,
+      "loss": 0.318,
+      "step": 2307
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6777367049105405e-05,
+      "loss": 0.257,
+      "step": 2308
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6774286010616484e-05,
+      "loss": 0.1981,
+      "step": 2309
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.677120367743229e-05,
+      "loss": 0.2112,
+      "step": 2310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.676812004989176e-05,
+      "loss": 0.2435,
+      "step": 2311
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6765035128333963e-05,
+      "loss": 0.3442,
+      "step": 2312
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.676194891309812e-05,
+      "loss": 0.2549,
+      "step": 2313
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.675886140452358e-05,
+      "loss": 0.1613,
+      "step": 2314
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6755772602949862e-05,
+      "loss": 0.2993,
+      "step": 2315
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6752682508716593e-05,
+      "loss": 0.2791,
+      "step": 2316
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.674959112216357e-05,
+      "loss": 0.2814,
+      "step": 2317
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6746498443630708e-05,
+      "loss": 0.2385,
+      "step": 2318
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6743404473458088e-05,
+      "loss": 0.2599,
+      "step": 2319
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6740309211985916e-05,
+      "loss": 0.2121,
+      "step": 2320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6737212659554546e-05,
+      "loss": 0.1642,
+      "step": 2321
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6734114816504478e-05,
+      "loss": 0.2753,
+      "step": 2322
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6731015683176338e-05,
+      "loss": 0.2681,
+      "step": 2323
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6727915259910918e-05,
+      "loss": 0.2943,
+      "step": 2324
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6724813547049135e-05,
+      "loss": 0.2911,
+      "step": 2325
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.672171054493205e-05,
+      "loss": 0.1813,
+      "step": 2326
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.671860625390087e-05,
+      "loss": 0.2528,
+      "step": 2327
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6715500674296938e-05,
+      "loss": 0.2287,
+      "step": 2328
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6712393806461742e-05,
+      "loss": 0.2866,
+      "step": 2329
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.670928565073692e-05,
+      "loss": 0.3535,
+      "step": 2330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6706176207464233e-05,
+      "loss": 0.2433,
+      "step": 2331
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.67030654769856e-05,
+      "loss": 0.2669,
+      "step": 2332
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6699953459643076e-05,
+      "loss": 0.2035,
+      "step": 2333
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6696840155778854e-05,
+      "loss": 0.29,
+      "step": 2334
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.669372556573527e-05,
+      "loss": 0.2839,
+      "step": 2335
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6690609689854813e-05,
+      "loss": 0.2247,
+      "step": 2336
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.668749252848009e-05,
+      "loss": 0.3246,
+      "step": 2337
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6684374081953874e-05,
+      "loss": 0.2012,
+      "step": 2338
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6681254350619055e-05,
+      "loss": 0.2285,
+      "step": 2339
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6678133334818688e-05,
+      "loss": 0.2879,
+      "step": 2340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6675011034895956e-05,
+      "loss": 0.2348,
+      "step": 2341
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.667188745119419e-05,
+      "loss": 0.2944,
+      "step": 2342
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6668762584056848e-05,
+      "loss": 0.284,
+      "step": 2343
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6665636433827547e-05,
+      "loss": 0.2949,
+      "step": 2344
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.666250900085003e-05,
+      "loss": 0.2584,
+      "step": 2345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.66593802854682e-05,
+      "loss": 0.3558,
+      "step": 2346
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6656250288026074e-05,
+      "loss": 0.3152,
+      "step": 2347
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6653119008867835e-05,
+      "loss": 0.2514,
+      "step": 2348
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.664998644833779e-05,
+      "loss": 0.2628,
+      "step": 2349
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6646852606780406e-05,
+      "loss": 0.2722,
+      "step": 2350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.664371748454027e-05,
+      "loss": 0.2554,
+      "step": 2351
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6640581081962118e-05,
+      "loss": 0.1964,
+      "step": 2352
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6637443399390832e-05,
+      "loss": 0.2481,
+      "step": 2353
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6634304437171425e-05,
+      "loss": 0.2916,
+      "step": 2354
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.663116419564906e-05,
+      "loss": 0.2547,
+      "step": 2355
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6628022675169034e-05,
+      "loss": 0.3314,
+      "step": 2356
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.662487987607679e-05,
+      "loss": 0.2227,
+      "step": 2357
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6621735798717912e-05,
+      "loss": 0.3799,
+      "step": 2358
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6618590443438108e-05,
+      "loss": 0.2556,
+      "step": 2359
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6615443810583253e-05,
+      "loss": 0.2761,
+      "step": 2360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6612295900499343e-05,
+      "loss": 0.3346,
+      "step": 2361
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6609146713532523e-05,
+      "loss": 0.1844,
+      "step": 2362
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.660599625002908e-05,
+      "loss": 0.1544,
+      "step": 2363
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6602844510335432e-05,
+      "loss": 0.1952,
+      "step": 2364
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6599691494798136e-05,
+      "loss": 0.2644,
+      "step": 2365
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6596537203763916e-05,
+      "loss": 0.3026,
+      "step": 2366
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.65933816375796e-05,
+      "loss": 0.155,
+      "step": 2367
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6590224796592177e-05,
+      "loss": 0.2708,
+      "step": 2368
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6587066681148773e-05,
+      "loss": 0.3088,
+      "step": 2369
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.658390729159665e-05,
+      "loss": 0.2399,
+      "step": 2370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6580746628283216e-05,
+      "loss": 0.2236,
+      "step": 2371
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.657758469155602e-05,
+      "loss": 0.2576,
+      "step": 2372
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.657442148176274e-05,
+      "loss": 0.2625,
+      "step": 2373
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6571256999251206e-05,
+      "loss": 0.3424,
+      "step": 2374
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6568091244369377e-05,
+      "loss": 0.1656,
+      "step": 2375
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6564924217465365e-05,
+      "loss": 0.2096,
+      "step": 2376
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.656175591888741e-05,
+      "loss": 0.2891,
+      "step": 2377
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.65585863489839e-05,
+      "loss": 0.2385,
+      "step": 2378
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6555415508103356e-05,
+      "loss": 0.2955,
+      "step": 2379
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6552243396594447e-05,
+      "loss": 0.2483,
+      "step": 2380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6549070014805974e-05,
+      "loss": 0.2083,
+      "step": 2381
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.654589536308688e-05,
+      "loss": 0.2686,
+      "step": 2382
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6542719441786247e-05,
+      "loss": 0.2703,
+      "step": 2383
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6539542251253303e-05,
+      "loss": 0.2626,
+      "step": 2384
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6536363791837412e-05,
+      "loss": 0.3125,
+      "step": 2385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6533184063888065e-05,
+      "loss": 0.2361,
+      "step": 2386
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6530003067754914e-05,
+      "loss": 0.2026,
+      "step": 2387
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.652682080378773e-05,
+      "loss": 0.2858,
+      "step": 2388
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6523637272336447e-05,
+      "loss": 0.2251,
+      "step": 2389
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.652045247375111e-05,
+      "loss": 0.2749,
+      "step": 2390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6517266408381934e-05,
+      "loss": 0.207,
+      "step": 2391
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.651407907657924e-05,
+      "loss": 0.2422,
+      "step": 2392
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6510890478693518e-05,
+      "loss": 0.2425,
+      "step": 2393
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.650770061507538e-05,
+      "loss": 0.1857,
+      "step": 2394
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.650450948607558e-05,
+      "loss": 0.2445,
+      "step": 2395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.650131709204502e-05,
+      "loss": 0.3209,
+      "step": 2396
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6498123433334733e-05,
+      "loss": 0.2452,
+      "step": 2397
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6494928510295883e-05,
+      "loss": 0.2436,
+      "step": 2398
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.649173232327979e-05,
+      "loss": 0.2388,
+      "step": 2399
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6488534872637908e-05,
+      "loss": 0.2323,
+      "step": 2400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.648533615872182e-05,
+      "loss": 0.2609,
+      "step": 2401
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6482136181883263e-05,
+      "loss": 0.2173,
+      "step": 2402
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6478934942474097e-05,
+      "loss": 0.3699,
+      "step": 2403
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.647573244084633e-05,
+      "loss": 0.2141,
+      "step": 2404
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6472528677352117e-05,
+      "loss": 0.2667,
+      "step": 2405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.646932365234373e-05,
+      "loss": 0.2429,
+      "step": 2406
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6466117366173597e-05,
+      "loss": 0.3187,
+      "step": 2407
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6462909819194283e-05,
+      "loss": 0.2332,
+      "step": 2408
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6459701011758487e-05,
+      "loss": 0.3051,
+      "step": 2409
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6456490944219044e-05,
+      "loss": 0.2347,
+      "step": 2410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6453279616928934e-05,
+      "loss": 0.3601,
+      "step": 2411
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.645006703024128e-05,
+      "loss": 0.2608,
+      "step": 2412
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6446853184509326e-05,
+      "loss": 0.2427,
+      "step": 2413
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6443638080086466e-05,
+      "loss": 0.2399,
+      "step": 2414
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.644042171732624e-05,
+      "loss": 0.2937,
+      "step": 2415
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6437204096582314e-05,
+      "loss": 0.2357,
+      "step": 2416
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6433985218208488e-05,
+      "loss": 0.2481,
+      "step": 2417
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.643076508255872e-05,
+      "loss": 0.3493,
+      "step": 2418
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.642754368998709e-05,
+      "loss": 0.2955,
+      "step": 2419
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.642432104084782e-05,
+      "loss": 0.2862,
+      "step": 2420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6421097135495268e-05,
+      "loss": 0.2382,
+      "step": 2421
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.641787197428394e-05,
+      "loss": 0.2205,
+      "step": 2422
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6414645557568473e-05,
+      "loss": 0.2835,
+      "step": 2423
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6411417885703636e-05,
+      "loss": 0.2703,
+      "step": 2424
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6408188959044347e-05,
+      "loss": 0.3073,
+      "step": 2425
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.640495877794566e-05,
+      "loss": 0.2681,
+      "step": 2426
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6401727342762757e-05,
+      "loss": 0.2636,
+      "step": 2427
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.639849465385097e-05,
+      "loss": 0.3317,
+      "step": 2428
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6395260711565764e-05,
+      "loss": 0.2574,
+      "step": 2429
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6392025516262742e-05,
+      "loss": 0.187,
+      "step": 2430
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6388789068297636e-05,
+      "loss": 0.2593,
+      "step": 2431
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.638555136802634e-05,
+      "loss": 0.2438,
+      "step": 2432
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6382312415804858e-05,
+      "loss": 0.1854,
+      "step": 2433
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6379072211989348e-05,
+      "loss": 0.2556,
+      "step": 2434
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6375830756936104e-05,
+      "loss": 0.2315,
+      "step": 2435
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6372588051001547e-05,
+      "loss": 0.3022,
+      "step": 2436
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6369344094542252e-05,
+      "loss": 0.2503,
+      "step": 2437
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6366098887914917e-05,
+      "loss": 0.2478,
+      "step": 2438
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6362852431476388e-05,
+      "loss": 0.3701,
+      "step": 2439
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6359604725583644e-05,
+      "loss": 0.212,
+      "step": 2440
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6356355770593795e-05,
+      "loss": 0.2328,
+      "step": 2441
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.63531055668641e-05,
+      "loss": 0.2325,
+      "step": 2442
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6349854114751952e-05,
+      "loss": 0.3088,
+      "step": 2443
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.634660141461487e-05,
+      "loss": 0.1454,
+      "step": 2444
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.634334746681053e-05,
+      "loss": 0.2177,
+      "step": 2445
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.634009227169673e-05,
+      "loss": 0.1994,
+      "step": 2446
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.633683582963141e-05,
+      "loss": 0.2354,
+      "step": 2447
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6333578140972647e-05,
+      "loss": 0.1693,
+      "step": 2448
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6330319206078653e-05,
+      "loss": 0.3682,
+      "step": 2449
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.632705902530778e-05,
+      "loss": 0.1669,
+      "step": 2450
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.632379759901852e-05,
+      "loss": 0.3293,
+      "step": 2451
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6320534927569493e-05,
+      "loss": 0.2404,
+      "step": 2452
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6317271011319467e-05,
+      "loss": 0.2754,
+      "step": 2453
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.631400585062733e-05,
+      "loss": 0.2272,
+      "step": 2454
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.631073944585213e-05,
+      "loss": 0.2338,
+      "step": 2455
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6307471797353034e-05,
+      "loss": 0.2453,
+      "step": 2456
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6304202905489344e-05,
+      "loss": 0.2414,
+      "step": 2457
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6300932770620518e-05,
+      "loss": 0.2151,
+      "step": 2458
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6297661393106135e-05,
+      "loss": 0.384,
+      "step": 2459
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.629438877330591e-05,
+      "loss": 0.2205,
+      "step": 2460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6291114911579702e-05,
+      "loss": 0.2391,
+      "step": 2461
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.62878398082875e-05,
+      "loss": 0.2667,
+      "step": 2462
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.628456346378944e-05,
+      "loss": 0.2845,
+      "step": 2463
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6281285878445778e-05,
+      "loss": 0.3031,
+      "step": 2464
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.627800705261693e-05,
+      "loss": 0.2161,
+      "step": 2465
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.627472698666342e-05,
+      "loss": 0.2747,
+      "step": 2466
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6271445680945928e-05,
+      "loss": 0.343,
+      "step": 2467
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.626816313582526e-05,
+      "loss": 0.1522,
+      "step": 2468
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6264879351662373e-05,
+      "loss": 0.2188,
+      "step": 2469
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6261594328818346e-05,
+      "loss": 0.2833,
+      "step": 2470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6258308067654393e-05,
+      "loss": 0.2809,
+      "step": 2471
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6255020568531882e-05,
+      "loss": 0.2505,
+      "step": 2472
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.625173183181229e-05,
+      "loss": 0.2658,
+      "step": 2473
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6248441857857256e-05,
+      "loss": 0.2727,
+      "step": 2474
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6245150647028547e-05,
+      "loss": 0.2469,
+      "step": 2475
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.624185819968805e-05,
+      "loss": 0.2041,
+      "step": 2476
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.623856451619781e-05,
+      "loss": 0.2707,
+      "step": 2477
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6235269596919996e-05,
+      "loss": 0.2759,
+      "step": 2478
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6231973442216916e-05,
+      "loss": 0.2197,
+      "step": 2479
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6228676052451017e-05,
+      "loss": 0.2432,
+      "step": 2480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6225377427984876e-05,
+      "loss": 0.2639,
+      "step": 2481
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.622207756918121e-05,
+      "loss": 0.2429,
+      "step": 2482
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.621877647640286e-05,
+      "loss": 0.2373,
+      "step": 2483
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.621547415001283e-05,
+      "loss": 0.2885,
+      "step": 2484
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.621217059037423e-05,
+      "loss": 0.3256,
+      "step": 2485
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.620886579785032e-05,
+      "loss": 0.2399,
+      "step": 2486
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.62055597728045e-05,
+      "loss": 0.1818,
+      "step": 2487
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.620225251560029e-05,
+      "loss": 0.3361,
+      "step": 2488
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6198944026601356e-05,
+      "loss": 0.2233,
+      "step": 2489
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6195634306171507e-05,
+      "loss": 0.2346,
+      "step": 2490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.619232335467467e-05,
+      "loss": 0.3417,
+      "step": 2491
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.618901117247491e-05,
+      "loss": 0.2159,
+      "step": 2492
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.618569775993645e-05,
+      "loss": 0.2915,
+      "step": 2493
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.618238311742362e-05,
+      "loss": 0.3489,
+      "step": 2494
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6179067245300895e-05,
+      "loss": 0.2631,
+      "step": 2495
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6175750143932898e-05,
+      "loss": 0.2198,
+      "step": 2496
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.617243181368436e-05,
+      "loss": 0.2688,
+      "step": 2497
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6169112254920176e-05,
+      "loss": 0.2511,
+      "step": 2498
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.616579146800536e-05,
+      "loss": 0.165,
+      "step": 2499
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.616246945330506e-05,
+      "loss": 0.2999,
+      "step": 2500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.615914621118457e-05,
+      "loss": 0.1719,
+      "step": 2501
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6155821742009304e-05,
+      "loss": 0.3153,
+      "step": 2502
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6152496046144823e-05,
+      "loss": 0.25,
+      "step": 2503
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6149169123956826e-05,
+      "loss": 0.2227,
+      "step": 2504
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6145840975811133e-05,
+      "loss": 0.2365,
+      "step": 2505
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6142511602073705e-05,
+      "loss": 0.3147,
+      "step": 2506
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6139181003110637e-05,
+      "loss": 0.2178,
+      "step": 2507
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6135849179288166e-05,
+      "loss": 0.1991,
+      "step": 2508
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6132516130972654e-05,
+      "loss": 0.2362,
+      "step": 2509
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.61291818585306e-05,
+      "loss": 0.2642,
+      "step": 2510
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6125846362328645e-05,
+      "loss": 0.19,
+      "step": 2511
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6122509642733554e-05,
+      "loss": 0.2034,
+      "step": 2512
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6119171700112233e-05,
+      "loss": 0.2544,
+      "step": 2513
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6115832534831716e-05,
+      "loss": 0.2404,
+      "step": 2514
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6112492147259188e-05,
+      "loss": 0.2294,
+      "step": 2515
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6109150537761947e-05,
+      "loss": 0.2939,
+      "step": 2516
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.610580770670743e-05,
+      "loss": 0.2129,
+      "step": 2517
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6102463654463227e-05,
+      "loss": 0.3378,
+      "step": 2518
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6099118381397037e-05,
+      "loss": 0.2407,
+      "step": 2519
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6095771887876712e-05,
+      "loss": 0.1934,
+      "step": 2520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6092424174270226e-05,
+      "loss": 0.266,
+      "step": 2521
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6089075240945697e-05,
+      "loss": 0.2758,
+      "step": 2522
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.608572508827137e-05,
+      "loss": 0.4008,
+      "step": 2523
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.608237371661562e-05,
+      "loss": 0.2731,
+      "step": 2524
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.607902112634697e-05,
+      "loss": 0.2507,
+      "step": 2525
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.607566731783407e-05,
+      "loss": 0.2229,
+      "step": 2526
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6072312291445705e-05,
+      "loss": 0.2413,
+      "step": 2527
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.606895604755078e-05,
+      "loss": 0.3111,
+      "step": 2528
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6065598586518357e-05,
+      "loss": 0.2904,
+      "step": 2529
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.606223990871762e-05,
+      "loss": 0.2948,
+      "step": 2530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6058880014517878e-05,
+      "loss": 0.1211,
+      "step": 2531
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.60555189042886e-05,
+      "loss": 0.22,
+      "step": 2532
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6052156578399363e-05,
+      "loss": 0.2614,
+      "step": 2533
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6048793037219883e-05,
+      "loss": 0.2341,
+      "step": 2534
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.604542828112002e-05,
+      "loss": 0.2726,
+      "step": 2535
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6042062310469754e-05,
+      "loss": 0.2225,
+      "step": 2536
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6038695125639213e-05,
+      "loss": 0.2485,
+      "step": 2537
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6035326726998654e-05,
+      "loss": 0.223,
+      "step": 2538
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.603195711491845e-05,
+      "loss": 0.2191,
+      "step": 2539
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.602858628976914e-05,
+      "loss": 0.3343,
+      "step": 2540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6025214251921362e-05,
+      "loss": 0.2579,
+      "step": 2541
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6021841001745915e-05,
+      "loss": 0.2404,
+      "step": 2542
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6018466539613722e-05,
+      "loss": 0.3158,
+      "step": 2543
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.601509086589583e-05,
+      "loss": 0.2984,
+      "step": 2544
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.601171398096342e-05,
+      "loss": 0.2469,
+      "step": 2545
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6008335885187826e-05,
+      "loss": 0.1905,
+      "step": 2546
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.60049565789405e-05,
+      "loss": 0.2399,
+      "step": 2547
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6001576062593024e-05,
+      "loss": 0.2459,
+      "step": 2548
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.599819433651712e-05,
+      "loss": 0.3146,
+      "step": 2549
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5994811401084643e-05,
+      "loss": 0.2092,
+      "step": 2550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5991427256667575e-05,
+      "loss": 0.2296,
+      "step": 2551
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5988041903638043e-05,
+      "loss": 0.2287,
+      "step": 2552
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.598465534236829e-05,
+      "loss": 0.2905,
+      "step": 2553
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5981267573230704e-05,
+      "loss": 0.2709,
+      "step": 2554
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.59778785965978e-05,
+      "loss": 0.2518,
+      "step": 2555
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5974488412842233e-05,
+      "loss": 0.2134,
+      "step": 2556
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5971097022336785e-05,
+      "loss": 0.2734,
+      "step": 2557
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5967704425454373e-05,
+      "loss": 0.2343,
+      "step": 2558
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.596431062256804e-05,
+      "loss": 0.3213,
+      "step": 2559
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.596091561405097e-05,
+      "loss": 0.2322,
+      "step": 2560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.595751940027648e-05,
+      "loss": 0.2149,
+      "step": 2561
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5954121981618016e-05,
+      "loss": 0.3049,
+      "step": 2562
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.595072335844915e-05,
+      "loss": 0.2977,
+      "step": 2563
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5947323531143598e-05,
+      "loss": 0.2414,
+      "step": 2564
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5943922500075202e-05,
+      "loss": 0.2113,
+      "step": 2565
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5940520265617943e-05,
+      "loss": 0.226,
+      "step": 2566
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5937116828145918e-05,
+      "loss": 0.2392,
+      "step": 2567
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5933712188033377e-05,
+      "loss": 0.2021,
+      "step": 2568
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5930306345654692e-05,
+      "loss": 0.2879,
+      "step": 2569
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5926899301384363e-05,
+      "loss": 0.1609,
+      "step": 2570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.592349105559703e-05,
+      "loss": 0.2007,
+      "step": 2571
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.592008160866746e-05,
+      "loss": 0.3551,
+      "step": 2572
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.591667096097056e-05,
+      "loss": 0.2087,
+      "step": 2573
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.591325911288136e-05,
+      "loss": 0.2003,
+      "step": 2574
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5909846064775025e-05,
+      "loss": 0.233,
+      "step": 2575
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5906431817026847e-05,
+      "loss": 0.2024,
+      "step": 2576
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5903016370012265e-05,
+      "loss": 0.2136,
+      "step": 2577
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5899599724106836e-05,
+      "loss": 0.2314,
+      "step": 2578
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5896181879686253e-05,
+      "loss": 0.1838,
+      "step": 2579
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5892762837126334e-05,
+      "loss": 0.253,
+      "step": 2580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5889342596803048e-05,
+      "loss": 0.2784,
+      "step": 2581
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5885921159092472e-05,
+      "loss": 0.2279,
+      "step": 2582
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5882498524370833e-05,
+      "loss": 0.2073,
+      "step": 2583
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.587907469301448e-05,
+      "loss": 0.1896,
+      "step": 2584
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5875649665399892e-05,
+      "loss": 0.3025,
+      "step": 2585
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5872223441903688e-05,
+      "loss": 0.3221,
+      "step": 2586
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5868796022902617e-05,
+      "loss": 0.2764,
+      "step": 2587
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5865367408773548e-05,
+      "loss": 0.236,
+      "step": 2588
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.58619375998935e-05,
+      "loss": 0.2773,
+      "step": 2589
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5858506596639603e-05,
+      "loss": 0.229,
+      "step": 2590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5855074399389135e-05,
+      "loss": 0.2123,
+      "step": 2591
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5851641008519495e-05,
+      "loss": 0.2172,
+      "step": 2592
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5848206424408224e-05,
+      "loss": 0.262,
+      "step": 2593
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.584477064743298e-05,
+      "loss": 0.2416,
+      "step": 2594
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5841333677971564e-05,
+      "loss": 0.2624,
+      "step": 2595
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.58378955164019e-05,
+      "loss": 0.1447,
+      "step": 2596
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5834456163102053e-05,
+      "loss": 0.1543,
+      "step": 2597
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.58310156184502e-05,
+      "loss": 0.1245,
+      "step": 2598
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5827573882824676e-05,
+      "loss": 0.2048,
+      "step": 2599
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5824130956603924e-05,
+      "loss": 0.2338,
+      "step": 2600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5820686840166532e-05,
+      "loss": 0.1854,
+      "step": 2601
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.581724153389121e-05,
+      "loss": 0.1779,
+      "step": 2602
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5813795038156805e-05,
+      "loss": 0.2063,
+      "step": 2603
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5810347353342285e-05,
+      "loss": 0.1546,
+      "step": 2604
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.580689847982677e-05,
+      "loss": 0.2797,
+      "step": 2605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.580344841798948e-05,
+      "loss": 0.2275,
+      "step": 2606
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5799997168209795e-05,
+      "loss": 0.2279,
+      "step": 2607
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5796544730867206e-05,
+      "loss": 0.2133,
+      "step": 2608
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.579309110634135e-05,
+      "loss": 0.2104,
+      "step": 2609
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.578963629501197e-05,
+      "loss": 0.1906,
+      "step": 2610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.578618029725897e-05,
+      "loss": 0.2542,
+      "step": 2611
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5782723113462366e-05,
+      "loss": 0.27,
+      "step": 2612
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5779264744002306e-05,
+      "loss": 0.317,
+      "step": 2613
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5775805189259075e-05,
+      "loss": 0.2156,
+      "step": 2614
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.577234444961308e-05,
+      "loss": 0.2538,
+      "step": 2615
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.576888252544486e-05,
+      "loss": 0.3047,
+      "step": 2616
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5765419417135094e-05,
+      "loss": 0.2356,
+      "step": 2617
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.576195512506458e-05,
+      "loss": 0.2625,
+      "step": 2618
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.575848964961425e-05,
+      "loss": 0.2736,
+      "step": 2619
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5755022991165168e-05,
+      "loss": 0.2086,
+      "step": 2620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5751555150098524e-05,
+      "loss": 0.2162,
+      "step": 2621
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.574808612679564e-05,
+      "loss": 0.2251,
+      "step": 2622
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.574461592163797e-05,
+      "loss": 0.318,
+      "step": 2623
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5741144535007097e-05,
+      "loss": 0.2357,
+      "step": 2624
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5737671967284734e-05,
+      "loss": 0.2094,
+      "step": 2625
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5734198218852717e-05,
+      "loss": 0.2624,
+      "step": 2626
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5730723290093016e-05,
+      "loss": 0.2932,
+      "step": 2627
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5727247181387747e-05,
+      "loss": 0.2265,
+      "step": 2628
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.572376989311913e-05,
+      "loss": 0.2404,
+      "step": 2629
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.572029142566953e-05,
+      "loss": 0.2516,
+      "step": 2630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.571681177942143e-05,
+      "loss": 0.2681,
+      "step": 2631
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5713330954757458e-05,
+      "loss": 0.2432,
+      "step": 2632
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5709848952060368e-05,
+      "loss": 0.2232,
+      "step": 2633
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5706365771713032e-05,
+      "loss": 0.2321,
+      "step": 2634
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.570288141409846e-05,
+      "loss": 0.2695,
+      "step": 2635
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5699395879599788e-05,
+      "loss": 0.2374,
+      "step": 2636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5695909168600296e-05,
+      "loss": 0.4287,
+      "step": 2637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5692421281483366e-05,
+      "loss": 0.215,
+      "step": 2638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5688932218632534e-05,
+      "loss": 0.1486,
+      "step": 2639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.568544198043145e-05,
+      "loss": 0.2441,
+      "step": 2640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5681950567263907e-05,
+      "loss": 0.359,
+      "step": 2641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.567845797951381e-05,
+      "loss": 0.2938,
+      "step": 2642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.567496421756521e-05,
+      "loss": 0.3615,
+      "step": 2643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.567146928180227e-05,
+      "loss": 0.2192,
+      "step": 2644
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5667973172609304e-05,
+      "loss": 0.2497,
+      "step": 2645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.566447589037073e-05,
+      "loss": 0.2057,
+      "step": 2646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.566097743547112e-05,
+      "loss": 0.2912,
+      "step": 2647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5657477808295152e-05,
+      "loss": 0.2302,
+      "step": 2648
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5653977009227647e-05,
+      "loss": 0.2459,
+      "step": 2649
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5650475038653555e-05,
+      "loss": 0.3317,
+      "step": 2650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5646971896957948e-05,
+      "loss": 0.287,
+      "step": 2651
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5643467584526028e-05,
+      "loss": 0.2366,
+      "step": 2652
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5639962101743136e-05,
+      "loss": 0.2575,
+      "step": 2653
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.563645544899472e-05,
+      "loss": 0.2639,
+      "step": 2654
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.563294762666638e-05,
+      "loss": 0.2358,
+      "step": 2655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5629438635143833e-05,
+      "loss": 0.1917,
+      "step": 2656
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5625928474812922e-05,
+      "loss": 0.2367,
+      "step": 2657
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.562241714605963e-05,
+      "loss": 0.2682,
+      "step": 2658
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5618904649270053e-05,
+      "loss": 0.1935,
+      "step": 2659
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.561539098483043e-05,
+      "loss": 0.2893,
+      "step": 2660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5611876153127117e-05,
+      "loss": 0.2327,
+      "step": 2661
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5608360154546607e-05,
+      "loss": 0.2923,
+      "step": 2662
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5604842989475514e-05,
+      "loss": 0.2834,
+      "step": 2663
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.560132465830059e-05,
+      "loss": 0.2534,
+      "step": 2664
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5597805161408702e-05,
+      "loss": 0.2423,
+      "step": 2665
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5594284499186856e-05,
+      "loss": 0.1669,
+      "step": 2666
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5590762672022183e-05,
+      "loss": 0.2481,
+      "step": 2667
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.558723968030194e-05,
+      "loss": 0.3171,
+      "step": 2668
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5583715524413515e-05,
+      "loss": 0.2748,
+      "step": 2669
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5580190204744423e-05,
+      "loss": 0.1666,
+      "step": 2670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5576663721682304e-05,
+      "loss": 0.2561,
+      "step": 2671
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5573136075614932e-05,
+      "loss": 0.243,
+      "step": 2672
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5569607266930198e-05,
+      "loss": 0.2396,
+      "step": 2673
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5566077296016137e-05,
+      "loss": 0.3036,
+      "step": 2674
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5562546163260902e-05,
+      "loss": 0.2569,
+      "step": 2675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.555901386905277e-05,
+      "loss": 0.2397,
+      "step": 2676
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5555480413780153e-05,
+      "loss": 0.2093,
+      "step": 2677
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5551945797831587e-05,
+      "loss": 0.3019,
+      "step": 2678
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.554841002159574e-05,
+      "loss": 0.2042,
+      "step": 2679
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.55448730854614e-05,
+      "loss": 0.3179,
+      "step": 2680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.554133498981749e-05,
+      "loss": 0.2375,
+      "step": 2681
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.553779573505306e-05,
+      "loss": 0.176,
+      "step": 2682
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5534255321557282e-05,
+      "loss": 0.2311,
+      "step": 2683
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5530713749719455e-05,
+      "loss": 0.2491,
+      "step": 2684
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.552717101992901e-05,
+      "loss": 0.2631,
+      "step": 2685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5523627132575504e-05,
+      "loss": 0.1594,
+      "step": 2686
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.552008208804863e-05,
+      "loss": 0.2339,
+      "step": 2687
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5516535886738187e-05,
+      "loss": 0.1621,
+      "step": 2688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.551298852903412e-05,
+      "loss": 0.2078,
+      "step": 2689
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5509440015326498e-05,
+      "loss": 0.3126,
+      "step": 2690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5505890346005504e-05,
+      "loss": 0.153,
+      "step": 2691
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5502339521461466e-05,
+      "loss": 0.2122,
+      "step": 2692
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5498787542084828e-05,
+      "loss": 0.2412,
+      "step": 2693
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.549523440826617e-05,
+      "loss": 0.2173,
+      "step": 2694
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5491680120396186e-05,
+      "loss": 0.2423,
+      "step": 2695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5488124678865702e-05,
+      "loss": 0.2193,
+      "step": 2696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.548456808406568e-05,
+      "loss": 0.2753,
+      "step": 2697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5481010336387198e-05,
+      "loss": 0.215,
+      "step": 2698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5477451436221467e-05,
+      "loss": 0.2567,
+      "step": 2699
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5473891383959817e-05,
+      "loss": 0.2756,
+      "step": 2700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5470330179993716e-05,
+      "loss": 0.2946,
+      "step": 2701
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.546676782471474e-05,
+      "loss": 0.3029,
+      "step": 2702
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.546320431851463e-05,
+      "loss": 0.2357,
+      "step": 2703
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5459639661785194e-05,
+      "loss": 0.252,
+      "step": 2704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.545607385491843e-05,
+      "loss": 0.2421,
+      "step": 2705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.545250689830641e-05,
+      "loss": 0.291,
+      "step": 2706
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.544893879234137e-05,
+      "loss": 0.2234,
+      "step": 2707
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5445369537415643e-05,
+      "loss": 0.2035,
+      "step": 2708
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5441799133921718e-05,
+      "loss": 0.2042,
+      "step": 2709
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5438227582252186e-05,
+      "loss": 0.2341,
+      "step": 2710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.543465488279978e-05,
+      "loss": 0.2155,
+      "step": 2711
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5431081035957344e-05,
+      "loss": 0.277,
+      "step": 2712
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.542750604211786e-05,
+      "loss": 0.2008,
+      "step": 2713
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5423929901674434e-05,
+      "loss": 0.2133,
+      "step": 2714
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.54203526150203e-05,
+      "loss": 0.1926,
+      "step": 2715
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5416774182548806e-05,
+      "loss": 0.2884,
+      "step": 2716
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5413194604653442e-05,
+      "loss": 0.1972,
+      "step": 2717
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5409613881727817e-05,
+      "loss": 0.2027,
+      "step": 2718
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5406032014165658e-05,
+      "loss": 0.2675,
+      "step": 2719
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5402449002360837e-05,
+      "loss": 0.2949,
+      "step": 2720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.539886484670733e-05,
+      "loss": 0.2479,
+      "step": 2721
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5395279547599255e-05,
+      "loss": 0.1855,
+      "step": 2722
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5391693105430845e-05,
+      "loss": 0.2865,
+      "step": 2723
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.538810552059647e-05,
+      "loss": 0.238,
+      "step": 2724
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5384516793490617e-05,
+      "loss": 0.1962,
+      "step": 2725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.53809269245079e-05,
+      "loss": 0.2827,
+      "step": 2726
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.537733591404306e-05,
+      "loss": 0.2349,
+      "step": 2727
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.537374376249096e-05,
+      "loss": 0.1767,
+      "step": 2728
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5370150470246593e-05,
+      "loss": 0.23,
+      "step": 2729
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5366556037705077e-05,
+      "loss": 0.1924,
+      "step": 2730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5362960465261654e-05,
+      "loss": 0.1965,
+      "step": 2731
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5359363753311698e-05,
+      "loss": 0.2372,
+      "step": 2732
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5355765902250686e-05,
+      "loss": 0.1824,
+      "step": 2733
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.535216691247425e-05,
+      "loss": 0.2979,
+      "step": 2734
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5348566784378127e-05,
+      "loss": 0.1718,
+      "step": 2735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5344965518358193e-05,
+      "loss": 0.2556,
+      "step": 2736
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5341363114810427e-05,
+      "loss": 0.2244,
+      "step": 2737
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5337759574130957e-05,
+      "loss": 0.3159,
+      "step": 2738
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5334154896716027e-05,
+      "loss": 0.2234,
+      "step": 2739
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.533054908296201e-05,
+      "loss": 0.2791,
+      "step": 2740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5326942133265392e-05,
+      "loss": 0.2591,
+      "step": 2741
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5323334048022792e-05,
+      "loss": 0.2266,
+      "step": 2742
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.531972482763096e-05,
+      "loss": 0.2135,
+      "step": 2743
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5316114472486757e-05,
+      "loss": 0.345,
+      "step": 2744
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5312502982987177e-05,
+      "loss": 0.2316,
+      "step": 2745
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5308890359529342e-05,
+      "loss": 0.2522,
+      "step": 2746
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5305276602510494e-05,
+      "loss": 0.256,
+      "step": 2747
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5301661712328e-05,
+      "loss": 0.1708,
+      "step": 2748
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5298045689379347e-05,
+      "loss": 0.253,
+      "step": 2749
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5294428534062147e-05,
+      "loss": 0.2149,
+      "step": 2750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.529081024677416e-05,
+      "loss": 0.2842,
+      "step": 2751
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5287190827913232e-05,
+      "loss": 0.3135,
+      "step": 2752
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5283570277877362e-05,
+      "loss": 0.2191,
+      "step": 2753
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5279948597064664e-05,
+      "loss": 0.2311,
+      "step": 2754
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5276325785873373e-05,
+      "loss": 0.2816,
+      "step": 2755
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.527270184470185e-05,
+      "loss": 0.3387,
+      "step": 2756
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5269076773948593e-05,
+      "loss": 0.3,
+      "step": 2757
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.52654505740122e-05,
+      "loss": 0.2224,
+      "step": 2758
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5261823245291413e-05,
+      "loss": 0.1924,
+      "step": 2759
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.525819478818509e-05,
+      "loss": 0.2357,
+      "step": 2760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.525456520309222e-05,
+      "loss": 0.2458,
+      "step": 2761
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5250934490411895e-05,
+      "loss": 0.2421,
+      "step": 2762
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5247302650543364e-05,
+      "loss": 0.1972,
+      "step": 2763
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5243669683885974e-05,
+      "loss": 0.2065,
+      "step": 2764
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5240035590839208e-05,
+      "loss": 0.2298,
+      "step": 2765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.523640037180266e-05,
+      "loss": 0.1776,
+      "step": 2766
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.523276402717607e-05,
+      "loss": 0.1558,
+      "step": 2767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5229126557359284e-05,
+      "loss": 0.2479,
+      "step": 2768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5225487962752273e-05,
+      "loss": 0.243,
+      "step": 2769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5221848243755138e-05,
+      "loss": 0.2273,
+      "step": 2770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5218207400768097e-05,
+      "loss": 0.2715,
+      "step": 2771
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5214565434191503e-05,
+      "loss": 0.2642,
+      "step": 2772
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5210922344425826e-05,
+      "loss": 0.3229,
+      "step": 2773
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5207278131871644e-05,
+      "loss": 0.196,
+      "step": 2774
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5203632796929686e-05,
+      "loss": 0.2106,
+      "step": 2775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.519998634000079e-05,
+      "loss": 0.1807,
+      "step": 2776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5196338761485918e-05,
+      "loss": 0.2154,
+      "step": 2777
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5192690061786157e-05,
+      "loss": 0.2811,
+      "step": 2778
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5189040241302707e-05,
+      "loss": 0.2187,
+      "step": 2779
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5185389300436912e-05,
+      "loss": 0.2598,
+      "step": 2780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5181737239590226e-05,
+      "loss": 0.1665,
+      "step": 2781
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5178084059164226e-05,
+      "loss": 0.1898,
+      "step": 2782
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.517442975956061e-05,
+      "loss": 0.243,
+      "step": 2783
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5170774341181207e-05,
+      "loss": 0.2139,
+      "step": 2784
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.516711780442797e-05,
+      "loss": 0.3291,
+      "step": 2785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.516346014970296e-05,
+      "loss": 0.2002,
+      "step": 2786
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5159801377408383e-05,
+      "loss": 0.2514,
+      "step": 2787
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5156141487946546e-05,
+      "loss": 0.2614,
+      "step": 2788
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5152480481719894e-05,
+      "loss": 0.2372,
+      "step": 2789
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.514881835913099e-05,
+      "loss": 0.2333,
+      "step": 2790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5145155120582507e-05,
+      "loss": 0.1966,
+      "step": 2791
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5141490766477274e-05,
+      "loss": 0.202,
+      "step": 2792
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.513782529721821e-05,
+      "loss": 0.2127,
+      "step": 2793
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5134158713208364e-05,
+      "loss": 0.2144,
+      "step": 2794
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.513049101485092e-05,
+      "loss": 0.1546,
+      "step": 2795
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5126822202549178e-05,
+      "loss": 0.2468,
+      "step": 2796
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5123152276706554e-05,
+      "loss": 0.2862,
+      "step": 2797
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.511948123772659e-05,
+      "loss": 0.299,
+      "step": 2798
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5115809086012952e-05,
+      "loss": 0.2539,
+      "step": 2799
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5112135821969438e-05,
+      "loss": 0.2195,
+      "step": 2800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5108461445999945e-05,
+      "loss": 0.2452,
+      "step": 2801
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5104785958508517e-05,
+      "loss": 0.2791,
+      "step": 2802
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5101109359899296e-05,
+      "loss": 0.2275,
+      "step": 2803
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.509743165057657e-05,
+      "loss": 0.3091,
+      "step": 2804
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5093752830944738e-05,
+      "loss": 0.3049,
+      "step": 2805
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5090072901408314e-05,
+      "loss": 0.2697,
+      "step": 2806
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5086391862371955e-05,
+      "loss": 0.1775,
+      "step": 2807
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5082709714240405e-05,
+      "loss": 0.2803,
+      "step": 2808
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5079026457418574e-05,
+      "loss": 0.1599,
+      "step": 2809
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5075342092311455e-05,
+      "loss": 0.21,
+      "step": 2810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5071656619324184e-05,
+      "loss": 0.17,
+      "step": 2811
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.506797003886202e-05,
+      "loss": 0.2192,
+      "step": 2812
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.506428235133033e-05,
+      "loss": 0.188,
+      "step": 2813
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5060593557134616e-05,
+      "loss": 0.3145,
+      "step": 2814
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.505690365668049e-05,
+      "loss": 0.2408,
+      "step": 2815
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5053212650373695e-05,
+      "loss": 0.2125,
+      "step": 2816
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.50495205386201e-05,
+      "loss": 0.2021,
+      "step": 2817
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5045827321825673e-05,
+      "loss": 0.2638,
+      "step": 2818
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.504213300039653e-05,
+      "loss": 0.1702,
+      "step": 2819
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.503843757473889e-05,
+      "loss": 0.2433,
+      "step": 2820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5034741045259103e-05,
+      "loss": 0.2228,
+      "step": 2821
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5031043412363644e-05,
+      "loss": 0.1394,
+      "step": 2822
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5027344676459096e-05,
+      "loss": 0.3157,
+      "step": 2823
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5023644837952167e-05,
+      "loss": 0.2585,
+      "step": 2824
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5019943897249697e-05,
+      "loss": 0.2933,
+      "step": 2825
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5016241854758638e-05,
+      "loss": 0.2222,
+      "step": 2826
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5012538710886064e-05,
+      "loss": 0.1322,
+      "step": 2827
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.500883446603917e-05,
+      "loss": 0.1572,
+      "step": 2828
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5005129120625276e-05,
+      "loss": 0.2547,
+      "step": 2829
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5001422675051816e-05,
+      "loss": 0.1882,
+      "step": 2830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4997715129726352e-05,
+      "loss": 0.2038,
+      "step": 2831
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4994006485056563e-05,
+      "loss": 0.2944,
+      "step": 2832
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4990296741450256e-05,
+      "loss": 0.3041,
+      "step": 2833
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4986585899315346e-05,
+      "loss": 0.16,
+      "step": 2834
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.498287395905987e-05,
+      "loss": 0.1438,
+      "step": 2835
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4979160921092003e-05,
+      "loss": 0.2236,
+      "step": 2836
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.497544678582002e-05,
+      "loss": 0.2293,
+      "step": 2837
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4971731553652337e-05,
+      "loss": 0.2442,
+      "step": 2838
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4968015224997463e-05,
+      "loss": 0.2999,
+      "step": 2839
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4964297800264063e-05,
+      "loss": 0.2312,
+      "step": 2840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4960579279860887e-05,
+      "loss": 0.3217,
+      "step": 2841
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4956859664196833e-05,
+      "loss": 0.2356,
+      "step": 2842
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4953138953680898e-05,
+      "loss": 0.1972,
+      "step": 2843
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4949417148722216e-05,
+      "loss": 0.2754,
+      "step": 2844
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4945694249730037e-05,
+      "loss": 0.2519,
+      "step": 2845
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4941970257113725e-05,
+      "loss": 0.2974,
+      "step": 2846
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.493824517128277e-05,
+      "loss": 0.1883,
+      "step": 2847
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4934518992646777e-05,
+      "loss": 0.3105,
+      "step": 2848
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.493079172161548e-05,
+      "loss": 0.2341,
+      "step": 2849
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4927063358598733e-05,
+      "loss": 0.2357,
+      "step": 2850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4923333904006493e-05,
+      "loss": 0.2729,
+      "step": 2851
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4919603358248857e-05,
+      "loss": 0.3131,
+      "step": 2852
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4915871721736033e-05,
+      "loss": 0.147,
+      "step": 2853
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4912138994878353e-05,
+      "loss": 0.2199,
+      "step": 2854
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.490840517808626e-05,
+      "loss": 0.3025,
+      "step": 2855
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.490467027177032e-05,
+      "loss": 0.2967,
+      "step": 2856
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.490093427634123e-05,
+      "loss": 0.2795,
+      "step": 2857
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.48971971922098e-05,
+      "loss": 0.2724,
+      "step": 2858
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4893459019786943e-05,
+      "loss": 0.1418,
+      "step": 2859
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4889719759483725e-05,
+      "loss": 0.177,
+      "step": 2860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4885979411711302e-05,
+      "loss": 0.1955,
+      "step": 2861
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.488223797688097e-05,
+      "loss": 0.2035,
+      "step": 2862
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.487849545540413e-05,
+      "loss": 0.3027,
+      "step": 2863
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4874751847692302e-05,
+      "loss": 0.2942,
+      "step": 2864
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.487100715415714e-05,
+      "loss": 0.1519,
+      "step": 2865
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.486726137521041e-05,
+      "loss": 0.3168,
+      "step": 2866
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4863514511263986e-05,
+      "loss": 0.2207,
+      "step": 2867
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.485976656272988e-05,
+      "loss": 0.3488,
+      "step": 2868
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4856017530020215e-05,
+      "loss": 0.1915,
+      "step": 2869
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4852267413547234e-05,
+      "loss": 0.3341,
+      "step": 2870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.484851621372329e-05,
+      "loss": 0.2806,
+      "step": 2871
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.484476393096087e-05,
+      "loss": 0.2249,
+      "step": 2872
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4841010565672572e-05,
+      "loss": 0.2419,
+      "step": 2873
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4837256118271112e-05,
+      "loss": 0.2135,
+      "step": 2874
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4833500589169334e-05,
+      "loss": 0.1447,
+      "step": 2875
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4829743978780182e-05,
+      "loss": 0.2413,
+      "step": 2876
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4825986287516743e-05,
+      "loss": 0.2663,
+      "step": 2877
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4822227515792207e-05,
+      "loss": 0.2159,
+      "step": 2878
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4818467664019888e-05,
+      "loss": 0.2164,
+      "step": 2879
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4814706732613214e-05,
+      "loss": 0.2348,
+      "step": 2880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.481094472198574e-05,
+      "loss": 0.1714,
+      "step": 2881
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4807181632551134e-05,
+      "loss": 0.14,
+      "step": 2882
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4803417464723177e-05,
+      "loss": 0.1986,
+      "step": 2883
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.479965221891578e-05,
+      "loss": 0.2264,
+      "step": 2884
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4795885895542978e-05,
+      "loss": 0.2601,
+      "step": 2885
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4792118495018897e-05,
+      "loss": 0.1835,
+      "step": 2886
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.478835001775781e-05,
+      "loss": 0.212,
+      "step": 2887
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.478458046417409e-05,
+      "loss": 0.2498,
+      "step": 2888
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.478080983468224e-05,
+      "loss": 0.2471,
+      "step": 2889
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4777038129696876e-05,
+      "loss": 0.1976,
+      "step": 2890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4773265349632737e-05,
+      "loss": 0.2632,
+      "step": 2891
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4769491494904667e-05,
+      "loss": 0.2393,
+      "step": 2892
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.476571656592764e-05,
+      "loss": 0.1452,
+      "step": 2893
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4761940563116757e-05,
+      "loss": 0.1684,
+      "step": 2894
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4758163486887208e-05,
+      "loss": 0.355,
+      "step": 2895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.475438533765433e-05,
+      "loss": 0.2231,
+      "step": 2896
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4750606115833564e-05,
+      "loss": 0.2543,
+      "step": 2897
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.474682582184047e-05,
+      "loss": 0.1772,
+      "step": 2898
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.474304445609073e-05,
+      "loss": 0.2255,
+      "step": 2899
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4739262019000136e-05,
+      "loss": 0.3417,
+      "step": 2900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.473547851098461e-05,
+      "loss": 0.2376,
+      "step": 2901
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.473169393246018e-05,
+      "loss": 0.2189,
+      "step": 2902
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4727908283843003e-05,
+      "loss": 0.2103,
+      "step": 2903
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4724121565549344e-05,
+      "loss": 0.1885,
+      "step": 2904
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4720333777995583e-05,
+      "loss": 0.2805,
+      "step": 2905
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4716544921598227e-05,
+      "loss": 0.274,
+      "step": 2906
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.47127549967739e-05,
+      "loss": 0.2964,
+      "step": 2907
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.470896400393934e-05,
+      "loss": 0.2413,
+      "step": 2908
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.47051719435114e-05,
+      "loss": 0.2155,
+      "step": 2909
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4701378815907055e-05,
+      "loss": 0.2088,
+      "step": 2910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4697584621543393e-05,
+      "loss": 0.2377,
+      "step": 2911
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4693789360837625e-05,
+      "loss": 0.2932,
+      "step": 2912
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4689993034207077e-05,
+      "loss": 0.2667,
+      "step": 2913
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4686195642069193e-05,
+      "loss": 0.311,
+      "step": 2914
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4682397184841523e-05,
+      "loss": 0.287,
+      "step": 2915
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.467859766294175e-05,
+      "loss": 0.2379,
+      "step": 2916
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4674797076787675e-05,
+      "loss": 0.1679,
+      "step": 2917
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.46709954267972e-05,
+      "loss": 0.2016,
+      "step": 2918
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4667192713388344e-05,
+      "loss": 0.2378,
+      "step": 2919
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4663388936979273e-05,
+      "loss": 0.2282,
+      "step": 2920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4659584097988234e-05,
+      "loss": 0.2616,
+      "step": 2921
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4655778196833606e-05,
+      "loss": 0.1176,
+      "step": 2922
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.465197123393389e-05,
+      "loss": 0.2449,
+      "step": 2923
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4648163209707693e-05,
+      "loss": 0.2072,
+      "step": 2924
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.464435412457375e-05,
+      "loss": 0.1558,
+      "step": 2925
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.46405439789509e-05,
+      "loss": 0.2242,
+      "step": 2926
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4636732773258107e-05,
+      "loss": 0.1616,
+      "step": 2927
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4632920507914447e-05,
+      "loss": 0.2021,
+      "step": 2928
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4629107183339123e-05,
+      "loss": 0.3198,
+      "step": 2929
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.462529279995144e-05,
+      "loss": 0.2754,
+      "step": 2930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4621477358170818e-05,
+      "loss": 0.3182,
+      "step": 2931
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4617660858416816e-05,
+      "loss": 0.2834,
+      "step": 2932
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4613843301109095e-05,
+      "loss": 0.2051,
+      "step": 2933
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4610024686667417e-05,
+      "loss": 0.2167,
+      "step": 2934
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.460620501551169e-05,
+      "loss": 0.1745,
+      "step": 2935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4602384288061915e-05,
+      "loss": 0.1908,
+      "step": 2936
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4598562504738217e-05,
+      "loss": 0.2164,
+      "step": 2937
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.459473966596084e-05,
+      "loss": 0.195,
+      "step": 2938
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.459091577215015e-05,
+      "loss": 0.1796,
+      "step": 2939
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4587090823726605e-05,
+      "loss": 0.2967,
+      "step": 2940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4583264821110805e-05,
+      "loss": 0.3003,
+      "step": 2941
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.457943776472345e-05,
+      "loss": 0.2099,
+      "step": 2942
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4575609654985367e-05,
+      "loss": 0.2738,
+      "step": 2943
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.457178049231749e-05,
+      "loss": 0.2228,
+      "step": 2944
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.456795027714087e-05,
+      "loss": 0.2362,
+      "step": 2945
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.456411900987668e-05,
+      "loss": 0.181,
+      "step": 2946
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4560286690946203e-05,
+      "loss": 0.2058,
+      "step": 2947
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4556453320770838e-05,
+      "loss": 0.2053,
+      "step": 2948
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4552618899772107e-05,
+      "loss": 0.2319,
+      "step": 2949
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4548783428371627e-05,
+      "loss": 0.358,
+      "step": 2950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.454494690699116e-05,
+      "loss": 0.2318,
+      "step": 2951
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.454110933605256e-05,
+      "loss": 0.2533,
+      "step": 2952
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.45372707159778e-05,
+      "loss": 0.2643,
+      "step": 2953
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4533431047188984e-05,
+      "loss": 0.2134,
+      "step": 2954
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.452959033010832e-05,
+      "loss": 0.2614,
+      "step": 2955
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4525748565158125e-05,
+      "loss": 0.3472,
+      "step": 2956
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4521905752760833e-05,
+      "loss": 0.2492,
+      "step": 2957
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4518061893339014e-05,
+      "loss": 0.3481,
+      "step": 2958
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4514216987315322e-05,
+      "loss": 0.1989,
+      "step": 2959
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.451037103511255e-05,
+      "loss": 0.1761,
+      "step": 2960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.450652403715359e-05,
+      "loss": 0.2362,
+      "step": 2961
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4502675993861464e-05,
+      "loss": 0.3092,
+      "step": 2962
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4498826905659297e-05,
+      "loss": 0.293,
+      "step": 2963
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4494976772970334e-05,
+      "loss": 0.2566,
+      "step": 2964
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4491125596217935e-05,
+      "loss": 0.1982,
+      "step": 2965
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.448727337582557e-05,
+      "loss": 0.1674,
+      "step": 2966
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4483420112216827e-05,
+      "loss": 0.2739,
+      "step": 2967
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4479565805815415e-05,
+      "loss": 0.2234,
+      "step": 2968
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.447571045704515e-05,
+      "loss": 0.1841,
+      "step": 2969
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4471854066329957e-05,
+      "loss": 0.2417,
+      "step": 2970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.446799663409389e-05,
+      "loss": 0.2925,
+      "step": 2971
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4464138160761116e-05,
+      "loss": 0.2429,
+      "step": 2972
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.44602786467559e-05,
+      "loss": 0.2484,
+      "step": 2973
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.445641809250264e-05,
+      "loss": 0.2447,
+      "step": 2974
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4452556498425835e-05,
+      "loss": 0.233,
+      "step": 2975
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.444869386495011e-05,
+      "loss": 0.2582,
+      "step": 2976
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4444830192500193e-05,
+      "loss": 0.2061,
+      "step": 2977
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4440965481500933e-05,
+      "loss": 0.2377,
+      "step": 2978
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4437099732377298e-05,
+      "loss": 0.1272,
+      "step": 2979
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4433232945554354e-05,
+      "loss": 0.2058,
+      "step": 2980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4429365121457296e-05,
+      "loss": 0.1901,
+      "step": 2981
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.442549626051143e-05,
+      "loss": 0.1471,
+      "step": 2982
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4421626363142175e-05,
+      "loss": 0.1765,
+      "step": 2983
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4417755429775057e-05,
+      "loss": 0.172,
+      "step": 2984
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4413883460835733e-05,
+      "loss": 0.2108,
+      "step": 2985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.441001045674995e-05,
+      "loss": 0.2542,
+      "step": 2986
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.440613641794359e-05,
+      "loss": 0.3007,
+      "step": 2987
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.440226134484264e-05,
+      "loss": 0.1278,
+      "step": 2988
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4398385237873197e-05,
+      "loss": 0.1328,
+      "step": 2989
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.439450809746148e-05,
+      "loss": 0.1581,
+      "step": 2990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4390629924033815e-05,
+      "loss": 0.1766,
+      "step": 2991
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4386750718016648e-05,
+      "loss": 0.2941,
+      "step": 2992
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4382870479836532e-05,
+      "loss": 0.2189,
+      "step": 2993
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4378989209920136e-05,
+      "loss": 0.2116,
+      "step": 2994
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4375106908694245e-05,
+      "loss": 0.1733,
+      "step": 2995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4371223576585753e-05,
+      "loss": 0.284,
+      "step": 2996
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4367339214021668e-05,
+      "loss": 0.1755,
+      "step": 2997
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4363453821429116e-05,
+      "loss": 0.2553,
+      "step": 2998
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.435956739923533e-05,
+      "loss": 0.2765,
+      "step": 2999
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4355679947867666e-05,
+      "loss": 0.1839,
+      "step": 3000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.435179146775358e-05,
+      "loss": 0.2089,
+      "step": 3001
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4347901959320646e-05,
+      "loss": 0.2363,
+      "step": 3002
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.434401142299656e-05,
+      "loss": 0.1655,
+      "step": 3003
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4340119859209117e-05,
+      "loss": 0.2181,
+      "step": 3004
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4336227268386235e-05,
+      "loss": 0.2557,
+      "step": 3005
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4332333650955934e-05,
+      "loss": 0.1797,
+      "step": 3006
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.432843900734637e-05,
+      "loss": 0.2062,
+      "step": 3007
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4324543337985785e-05,
+      "loss": 0.2917,
+      "step": 3008
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4320646643302548e-05,
+      "loss": 0.2743,
+      "step": 3009
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4316748923725135e-05,
+      "loss": 0.183,
+      "step": 3010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4312850179682138e-05,
+      "loss": 0.2723,
+      "step": 3011
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4308950411602268e-05,
+      "loss": 0.2088,
+      "step": 3012
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4305049619914333e-05,
+      "loss": 0.2459,
+      "step": 3013
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.430114780504727e-05,
+      "loss": 0.2908,
+      "step": 3014
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4297244967430114e-05,
+      "loss": 0.3032,
+      "step": 3015
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.429334110749202e-05,
+      "loss": 0.2641,
+      "step": 3016
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.428943622566226e-05,
+      "loss": 0.2443,
+      "step": 3017
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4285530322370208e-05,
+      "loss": 0.2427,
+      "step": 3018
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.428162339804536e-05,
+      "loss": 0.1716,
+      "step": 3019
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4277715453117315e-05,
+      "loss": 0.2206,
+      "step": 3020
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.427380648801579e-05,
+      "loss": 0.1696,
+      "step": 3021
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.426989650317062e-05,
+      "loss": 0.1323,
+      "step": 3022
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4265985499011727e-05,
+      "loss": 0.3781,
+      "step": 3023
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4262073475969188e-05,
+      "loss": 0.2938,
+      "step": 3024
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4258160434473148e-05,
+      "loss": 0.2509,
+      "step": 3025
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.425424637495389e-05,
+      "loss": 0.2203,
+      "step": 3026
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4250331297841808e-05,
+      "loss": 0.2209,
+      "step": 3027
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4246415203567394e-05,
+      "loss": 0.2466,
+      "step": 3028
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4242498092561258e-05,
+      "loss": 0.2454,
+      "step": 3029
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.423857996525413e-05,
+      "loss": 0.1548,
+      "step": 3030
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4234660822076847e-05,
+      "loss": 0.2399,
+      "step": 3031
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4230740663460353e-05,
+      "loss": 0.3356,
+      "step": 3032
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.42268194898357e-05,
+      "loss": 0.2113,
+      "step": 3033
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4222897301634074e-05,
+      "loss": 0.1531,
+      "step": 3034
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4218974099286744e-05,
+      "loss": 0.226,
+      "step": 3035
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.421504988322511e-05,
+      "loss": 0.1871,
+      "step": 3036
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.421112465388067e-05,
+      "loss": 0.1681,
+      "step": 3037
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4207198411685055e-05,
+      "loss": 0.2308,
+      "step": 3038
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4203271157069978e-05,
+      "loss": 0.2889,
+      "step": 3039
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.419934289046728e-05,
+      "loss": 0.1577,
+      "step": 3040
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.419541361230892e-05,
+      "loss": 0.1312,
+      "step": 3041
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4191483323026955e-05,
+      "loss": 0.2869,
+      "step": 3042
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4187552023053557e-05,
+      "loss": 0.2327,
+      "step": 3043
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.418361971282101e-05,
+      "loss": 0.2527,
+      "step": 3044
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.417968639276171e-05,
+      "loss": 0.2398,
+      "step": 3045
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4175752063308163e-05,
+      "loss": 0.127,
+      "step": 3046
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4171816724892987e-05,
+      "loss": 0.1936,
+      "step": 3047
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4167880377948904e-05,
+      "loss": 0.2114,
+      "step": 3048
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4163943022908767e-05,
+      "loss": 0.2751,
+      "step": 3049
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.416000466020551e-05,
+      "loss": 0.3708,
+      "step": 3050
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.41560652902722e-05,
+      "loss": 0.2236,
+      "step": 3051
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4152124913542016e-05,
+      "loss": 0.2076,
+      "step": 3052
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4148183530448224e-05,
+      "loss": 0.2227,
+      "step": 3053
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4144241141424233e-05,
+      "loss": 0.1924,
+      "step": 3054
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4140297746903534e-05,
+      "loss": 0.2814,
+      "step": 3055
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.413635334731975e-05,
+      "loss": 0.251,
+      "step": 3056
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4132407943106597e-05,
+      "loss": 0.2784,
+      "step": 3057
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.412846153469792e-05,
+      "loss": 0.2258,
+      "step": 3058
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4124514122527653e-05,
+      "loss": 0.238,
+      "step": 3059
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.412056570702986e-05,
+      "loss": 0.2358,
+      "step": 3060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4116616288638704e-05,
+      "loss": 0.2276,
+      "step": 3061
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.411266586778846e-05,
+      "loss": 0.1904,
+      "step": 3062
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.410871444491352e-05,
+      "loss": 0.2115,
+      "step": 3063
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.410476202044837e-05,
+      "loss": 0.2689,
+      "step": 3064
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4100808594827634e-05,
+      "loss": 0.1911,
+      "step": 3065
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.409685416848601e-05,
+      "loss": 0.2546,
+      "step": 3066
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.409289874185834e-05,
+      "loss": 0.2508,
+      "step": 3067
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.408894231537955e-05,
+      "loss": 0.2676,
+      "step": 3068
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.408498488948469e-05,
+      "loss": 0.2637,
+      "step": 3069
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.408102646460892e-05,
+      "loss": 0.2037,
+      "step": 3070
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4077067041187506e-05,
+      "loss": 0.1452,
+      "step": 3071
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4073106619655824e-05,
+      "loss": 0.2725,
+      "step": 3072
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4069145200449357e-05,
+      "loss": 0.2285,
+      "step": 3073
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4065182784003706e-05,
+      "loss": 0.194,
+      "step": 3074
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.406121937075457e-05,
+      "loss": 0.199,
+      "step": 3075
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.405725496113777e-05,
+      "loss": 0.2362,
+      "step": 3076
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4053289555589228e-05,
+      "loss": 0.2401,
+      "step": 3077
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4049323154544986e-05,
+      "loss": 0.195,
+      "step": 3078
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4045355758441177e-05,
+      "loss": 0.2259,
+      "step": 3079
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4041387367714056e-05,
+      "loss": 0.2618,
+      "step": 3080
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4037417982799983e-05,
+      "loss": 0.2442,
+      "step": 3081
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4033447604135443e-05,
+      "loss": 0.1827,
+      "step": 3082
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4029476232157e-05,
+      "loss": 0.4067,
+      "step": 3083
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4025503867301362e-05,
+      "loss": 0.2722,
+      "step": 3084
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4021530510005313e-05,
+      "loss": 0.2504,
+      "step": 3085
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.401755616070577e-05,
+      "loss": 0.2269,
+      "step": 3086
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4013580819839746e-05,
+      "loss": 0.2112,
+      "step": 3087
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.400960448784437e-05,
+      "loss": 0.2125,
+      "step": 3088
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.400562716515688e-05,
+      "loss": 0.1783,
+      "step": 3089
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.400164885221462e-05,
+      "loss": 0.1913,
+      "step": 3090
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3997669549455034e-05,
+      "loss": 0.3467,
+      "step": 3091
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3993689257315697e-05,
+      "loss": 0.2328,
+      "step": 3092
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3989707976234282e-05,
+      "loss": 0.2404,
+      "step": 3093
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3985725706648554e-05,
+      "loss": 0.1975,
+      "step": 3094
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3981742448996413e-05,
+      "loss": 0.2473,
+      "step": 3095
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3977758203715853e-05,
+      "loss": 0.2239,
+      "step": 3096
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.397377297124498e-05,
+      "loss": 0.1817,
+      "step": 3097
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.396978675202201e-05,
+      "loss": 0.2222,
+      "step": 3098
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.396579954648526e-05,
+      "loss": 0.1886,
+      "step": 3099
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3961811355073167e-05,
+      "loss": 0.1515,
+      "step": 3100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3957822178224273e-05,
+      "loss": 0.2123,
+      "step": 3101
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3953832016377218e-05,
+      "loss": 0.1949,
+      "step": 3102
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3949840869970767e-05,
+      "loss": 0.229,
+      "step": 3103
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.394584873944378e-05,
+      "loss": 0.2255,
+      "step": 3104
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.394185562523523e-05,
+      "loss": 0.2431,
+      "step": 3105
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3937861527784197e-05,
+      "loss": 0.3732,
+      "step": 3106
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3933866447529873e-05,
+      "loss": 0.2262,
+      "step": 3107
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3929870384911555e-05,
+      "loss": 0.2224,
+      "step": 3108
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3925873340368647e-05,
+      "loss": 0.2744,
+      "step": 3109
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3921875314340663e-05,
+      "loss": 0.2322,
+      "step": 3110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3917876307267224e-05,
+      "loss": 0.2021,
+      "step": 3111
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3913876319588058e-05,
+      "loss": 0.2321,
+      "step": 3112
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3909875351743003e-05,
+      "loss": 0.2551,
+      "step": 3113
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3905873404172007e-05,
+      "loss": 0.2473,
+      "step": 3114
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.390187047731511e-05,
+      "loss": 0.2905,
+      "step": 3115
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3897866571612487e-05,
+      "loss": 0.2818,
+      "step": 3116
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3893861687504404e-05,
+      "loss": 0.2429,
+      "step": 3117
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3889855825431225e-05,
+      "loss": 0.3237,
+      "step": 3118
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.388584898583344e-05,
+      "loss": 0.1441,
+      "step": 3119
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3881841169151637e-05,
+      "loss": 0.2151,
+      "step": 3120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.387783237582652e-05,
+      "loss": 0.2466,
+      "step": 3121
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3873822606298888e-05,
+      "loss": 0.219,
+      "step": 3122
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3869811861009655e-05,
+      "loss": 0.2541,
+      "step": 3123
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3865800140399842e-05,
+      "loss": 0.2287,
+      "step": 3124
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3861787444910578e-05,
+      "loss": 0.2561,
+      "step": 3125
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.385777377498309e-05,
+      "loss": 0.2234,
+      "step": 3126
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3853759131058724e-05,
+      "loss": 0.2205,
+      "step": 3127
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.384974351357893e-05,
+      "loss": 0.2042,
+      "step": 3128
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3845726922985262e-05,
+      "loss": 0.2603,
+      "step": 3129
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.384170935971938e-05,
+      "loss": 0.2043,
+      "step": 3130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3837690824223058e-05,
+      "loss": 0.2798,
+      "step": 3131
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3833671316938166e-05,
+      "loss": 0.1767,
+      "step": 3132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3829650838306695e-05,
+      "loss": 0.2577,
+      "step": 3133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.382562938877073e-05,
+      "loss": 0.2562,
+      "step": 3134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3821606968772472e-05,
+      "loss": 0.2354,
+      "step": 3135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3817583578754215e-05,
+      "loss": 0.2211,
+      "step": 3136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3813559219158384e-05,
+      "loss": 0.2502,
+      "step": 3137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3809533890427485e-05,
+      "loss": 0.2465,
+      "step": 3138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3805507593004138e-05,
+      "loss": 0.2227,
+      "step": 3139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3801480327331086e-05,
+      "loss": 0.2101,
+      "step": 3140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.379745209385115e-05,
+      "loss": 0.2205,
+      "step": 3141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3793422893007292e-05,
+      "loss": 0.2657,
+      "step": 3142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3789392725242546e-05,
+      "loss": 0.3586,
+      "step": 3143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.378536159100007e-05,
+      "loss": 0.2751,
+      "step": 3144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3781329490723125e-05,
+      "loss": 0.2639,
+      "step": 3145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.377729642485509e-05,
+      "loss": 0.2289,
+      "step": 3146
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.377326239383942e-05,
+      "loss": 0.1926,
+      "step": 3147
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3769227398119712e-05,
+      "loss": 0.2158,
+      "step": 3148
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3765191438139643e-05,
+      "loss": 0.2708,
+      "step": 3149
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.376115451434301e-05,
+      "loss": 0.2224,
+      "step": 3150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3757116627173708e-05,
+      "loss": 0.2535,
+      "step": 3151
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3753077777075746e-05,
+      "loss": 0.2358,
+      "step": 3152
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.374903796449323e-05,
+      "loss": 0.2578,
+      "step": 3153
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3744997189870376e-05,
+      "loss": 0.2091,
+      "step": 3154
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3740955453651506e-05,
+      "loss": 0.2175,
+      "step": 3155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3736912756281054e-05,
+      "loss": 0.2994,
+      "step": 3156
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3732869098203543e-05,
+      "loss": 0.3175,
+      "step": 3157
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.372882447986362e-05,
+      "loss": 0.2558,
+      "step": 3158
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.372477890170602e-05,
+      "loss": 0.2523,
+      "step": 3159
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3720732364175604e-05,
+      "loss": 0.2348,
+      "step": 3160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.371668486771732e-05,
+      "loss": 0.2732,
+      "step": 3161
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.371263641277623e-05,
+      "loss": 0.2421,
+      "step": 3162
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3708586999797508e-05,
+      "loss": 0.2006,
+      "step": 3163
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3704536629226416e-05,
+      "loss": 0.2843,
+      "step": 3164
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3700485301508334e-05,
+      "loss": 0.2811,
+      "step": 3165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.369643301708874e-05,
+      "loss": 0.2533,
+      "step": 3166
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.369237977641323e-05,
+      "loss": 0.2552,
+      "step": 3167
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3688325579927494e-05,
+      "loss": 0.2216,
+      "step": 3168
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3684270428077327e-05,
+      "loss": 0.2139,
+      "step": 3169
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3680214321308636e-05,
+      "loss": 0.1674,
+      "step": 3170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3676157260067424e-05,
+      "loss": 0.2672,
+      "step": 3171
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3672099244799803e-05,
+      "loss": 0.1842,
+      "step": 3172
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.366804027595199e-05,
+      "loss": 0.1501,
+      "step": 3173
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3663980353970316e-05,
+      "loss": 0.3039,
+      "step": 3174
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.36599194793012e-05,
+      "loss": 0.1905,
+      "step": 3175
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3655857652391173e-05,
+      "loss": 0.3147,
+      "step": 3176
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3651794873686878e-05,
+      "loss": 0.1677,
+      "step": 3177
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3647731143635055e-05,
+      "loss": 0.1439,
+      "step": 3178
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3643666462682548e-05,
+      "loss": 0.169,
+      "step": 3179
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3639600831276305e-05,
+      "loss": 0.1973,
+      "step": 3180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3635534249863383e-05,
+      "loss": 0.1476,
+      "step": 3181
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3631466718890946e-05,
+      "loss": 0.2666,
+      "step": 3182
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3627398238806253e-05,
+      "loss": 0.1924,
+      "step": 3183
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3623328810056674e-05,
+      "loss": 0.2018,
+      "step": 3184
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3619258433089682e-05,
+      "loss": 0.2642,
+      "step": 3185
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.361518710835285e-05,
+      "loss": 0.2532,
+      "step": 3186
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3611114836293867e-05,
+      "loss": 0.2809,
+      "step": 3187
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.360704161736051e-05,
+      "loss": 0.1822,
+      "step": 3188
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.360296745200067e-05,
+      "loss": 0.1799,
+      "step": 3189
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.359889234066234e-05,
+      "loss": 0.2218,
+      "step": 3190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3594816283793626e-05,
+      "loss": 0.1968,
+      "step": 3191
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.359073928184272e-05,
+      "loss": 0.2657,
+      "step": 3192
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.358666133525793e-05,
+      "loss": 0.182,
+      "step": 3193
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3582582444487665e-05,
+      "loss": 0.2736,
+      "step": 3194
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3578502609980437e-05,
+      "loss": 0.2196,
+      "step": 3195
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.357442183218487e-05,
+      "loss": 0.2263,
+      "step": 3196
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.357034011154967e-05,
+      "loss": 0.1881,
+      "step": 3197
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3566257448523675e-05,
+      "loss": 0.2233,
+      "step": 3198
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3562173843555808e-05,
+      "loss": 0.1101,
+      "step": 3199
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3558089297095103e-05,
+      "loss": 0.2311,
+      "step": 3200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3554003809590686e-05,
+      "loss": 0.1571,
+      "step": 3201
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3549917381491804e-05,
+      "loss": 0.2507,
+      "step": 3202
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3545830013247798e-05,
+      "loss": 0.2397,
+      "step": 3203
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.354174170530811e-05,
+      "loss": 0.3284,
+      "step": 3204
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3537652458122288e-05,
+      "loss": 0.2466,
+      "step": 3205
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3533562272139987e-05,
+      "loss": 0.1677,
+      "step": 3206
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3529471147810962e-05,
+      "loss": 0.2547,
+      "step": 3207
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.352537908558507e-05,
+      "loss": 0.2638,
+      "step": 3208
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.352128608591227e-05,
+      "loss": 0.2221,
+      "step": 3209
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3517192149242632e-05,
+      "loss": 0.1316,
+      "step": 3210
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3513097276026318e-05,
+      "loss": 0.2589,
+      "step": 3211
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3509001466713603e-05,
+      "loss": 0.1749,
+      "step": 3212
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3504904721754857e-05,
+      "loss": 0.2789,
+      "step": 3213
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.350080704160056e-05,
+      "loss": 0.2516,
+      "step": 3214
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3496708426701284e-05,
+      "loss": 0.2028,
+      "step": 3215
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3492608877507717e-05,
+      "loss": 0.2881,
+      "step": 3216
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.348850839447064e-05,
+      "loss": 0.1807,
+      "step": 3217
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3484406978040946e-05,
+      "loss": 0.2742,
+      "step": 3218
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3480304628669616e-05,
+      "loss": 0.2214,
+      "step": 3219
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3476201346807754e-05,
+      "loss": 0.1856,
+      "step": 3220
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3472097132906544e-05,
+      "loss": 0.2109,
+      "step": 3221
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.346799198741729e-05,
+      "loss": 0.181,
+      "step": 3222
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.346388591079139e-05,
+      "loss": 0.1487,
+      "step": 3223
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.345977890348035e-05,
+      "loss": 0.2388,
+      "step": 3224
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3455670965935767e-05,
+      "loss": 0.1659,
+      "step": 3225
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3451562098609353e-05,
+      "loss": 0.2726,
+      "step": 3226
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3447452301952918e-05,
+      "loss": 0.2321,
+      "step": 3227
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.344334157641837e-05,
+      "loss": 0.2795,
+      "step": 3228
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.343922992245773e-05,
+      "loss": 0.2554,
+      "step": 3229
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.34351173405231e-05,
+      "loss": 0.2112,
+      "step": 3230
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3431003831066713e-05,
+      "loss": 0.2029,
+      "step": 3231
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3426889394540873e-05,
+      "loss": 0.2846,
+      "step": 3232
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3422774031398018e-05,
+      "loss": 0.1608,
+      "step": 3233
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3418657742090663e-05,
+      "loss": 0.1282,
+      "step": 3234
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3414540527071432e-05,
+      "loss": 0.1656,
+      "step": 3235
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3410422386793054e-05,
+      "loss": 0.1365,
+      "step": 3236
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3406303321708356e-05,
+      "loss": 0.2363,
+      "step": 3237
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3402183332270277e-05,
+      "loss": 0.19,
+      "step": 3238
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3398062418931838e-05,
+      "loss": 0.1947,
+      "step": 3239
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3393940582146177e-05,
+      "loss": 0.1904,
+      "step": 3240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3389817822366535e-05,
+      "loss": 0.2544,
+      "step": 3241
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3385694140046238e-05,
+      "loss": 0.1712,
+      "step": 3242
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3381569535638735e-05,
+      "loss": 0.1557,
+      "step": 3243
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3377444009597555e-05,
+      "loss": 0.2333,
+      "step": 3244
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3373317562376353e-05,
+      "loss": 0.1697,
+      "step": 3245
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3369190194428857e-05,
+      "loss": 0.2489,
+      "step": 3246
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3365061906208916e-05,
+      "loss": 0.2336,
+      "step": 3247
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3360932698170477e-05,
+      "loss": 0.0962,
+      "step": 3248
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3356802570767587e-05,
+      "loss": 0.1796,
+      "step": 3249
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3352671524454386e-05,
+      "loss": 0.2272,
+      "step": 3250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.334853955968513e-05,
+      "loss": 0.2455,
+      "step": 3251
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.334440667691416e-05,
+      "loss": 0.1619,
+      "step": 3252
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3340272876595933e-05,
+      "loss": 0.1885,
+      "step": 3253
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3336138159184998e-05,
+      "loss": 0.2482,
+      "step": 3254
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3332002525136004e-05,
+      "loss": 0.2887,
+      "step": 3255
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3327865974903706e-05,
+      "loss": 0.2181,
+      "step": 3256
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3323728508942953e-05,
+      "loss": 0.2294,
+      "step": 3257
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.331959012770871e-05,
+      "loss": 0.2793,
+      "step": 3258
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3315450831656017e-05,
+      "loss": 0.2,
+      "step": 3259
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.331131062124004e-05,
+      "loss": 0.1731,
+      "step": 3260
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3307169496916027e-05,
+      "loss": 0.2641,
+      "step": 3261
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3303027459139343e-05,
+      "loss": 0.2403,
+      "step": 3262
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3298884508365435e-05,
+      "loss": 0.1909,
+      "step": 3263
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.329474064504987e-05,
+      "loss": 0.2517,
+      "step": 3264
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3290595869648297e-05,
+      "loss": 0.1579,
+      "step": 3265
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.328645018261648e-05,
+      "loss": 0.2771,
+      "step": 3266
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3282303584410267e-05,
+      "loss": 0.1899,
+      "step": 3267
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.327815607548563e-05,
+      "loss": 0.1942,
+      "step": 3268
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3274007656298614e-05,
+      "loss": 0.2357,
+      "step": 3269
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3269858327305387e-05,
+      "loss": 0.2332,
+      "step": 3270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3265708088962202e-05,
+      "loss": 0.1996,
+      "step": 3271
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.326155694172542e-05,
+      "loss": 0.2419,
+      "step": 3272
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3257404886051496e-05,
+      "loss": 0.3113,
+      "step": 3273
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3253251922396995e-05,
+      "loss": 0.2501,
+      "step": 3274
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.324909805121857e-05,
+      "loss": 0.2153,
+      "step": 3275
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3244943272972975e-05,
+      "loss": 0.2649,
+      "step": 3276
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3240787588117077e-05,
+      "loss": 0.2803,
+      "step": 3277
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3236630997107828e-05,
+      "loss": 0.1354,
+      "step": 3278
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.323247350040228e-05,
+      "loss": 0.2316,
+      "step": 3279
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.32283150984576e-05,
+      "loss": 0.2353,
+      "step": 3280
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3224155791731037e-05,
+      "loss": 0.1461,
+      "step": 3281
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3219995580679945e-05,
+      "loss": 0.213,
+      "step": 3282
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3215834465761786e-05,
+      "loss": 0.1592,
+      "step": 3283
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3211672447434113e-05,
+      "loss": 0.1509,
+      "step": 3284
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3207509526154575e-05,
+      "loss": 0.1817,
+      "step": 3285
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3203345702380925e-05,
+      "loss": 0.2575,
+      "step": 3286
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.319918097657102e-05,
+      "loss": 0.2743,
+      "step": 3287
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3195015349182805e-05,
+      "loss": 0.1745,
+      "step": 3288
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.319084882067434e-05,
+      "loss": 0.1968,
+      "step": 3289
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3186681391503765e-05,
+      "loss": 0.2269,
+      "step": 3290
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3182513062129334e-05,
+      "loss": 0.1838,
+      "step": 3291
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3178343833009398e-05,
+      "loss": 0.2026,
+      "step": 3292
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3174173704602394e-05,
+      "loss": 0.2319,
+      "step": 3293
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.317000267736687e-05,
+      "loss": 0.2239,
+      "step": 3294
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3165830751761476e-05,
+      "loss": 0.2881,
+      "step": 3295
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3161657928244953e-05,
+      "loss": 0.2254,
+      "step": 3296
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.315748420727614e-05,
+      "loss": 0.1998,
+      "step": 3297
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.315330958931398e-05,
+      "loss": 0.2709,
+      "step": 3298
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.314913407481751e-05,
+      "loss": 0.2435,
+      "step": 3299
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.314495766424587e-05,
+      "loss": 0.188,
+      "step": 3300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3140780358058295e-05,
+      "loss": 0.1938,
+      "step": 3301
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.313660215671412e-05,
+      "loss": 0.1991,
+      "step": 3302
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.313242306067278e-05,
+      "loss": 0.2342,
+      "step": 3303
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3128243070393802e-05,
+      "loss": 0.2267,
+      "step": 3304
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3124062186336817e-05,
+      "loss": 0.1899,
+      "step": 3305
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.311988040896156e-05,
+      "loss": 0.197,
+      "step": 3306
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3115697738727847e-05,
+      "loss": 0.2584,
+      "step": 3307
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.311151417609561e-05,
+      "loss": 0.2802,
+      "step": 3308
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.310732972152487e-05,
+      "loss": 0.2009,
+      "step": 3309
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.310314437547575e-05,
+      "loss": 0.3331,
+      "step": 3310
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3098958138408462e-05,
+      "loss": 0.2512,
+      "step": 3311
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3094771010783328e-05,
+      "loss": 0.2899,
+      "step": 3312
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3090582993060756e-05,
+      "loss": 0.2514,
+      "step": 3313
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3086394085701272e-05,
+      "loss": 0.3036,
+      "step": 3314
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3082204289165473e-05,
+      "loss": 0.1838,
+      "step": 3315
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.307801360391408e-05,
+      "loss": 0.2254,
+      "step": 3316
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3073822030407883e-05,
+      "loss": 0.2433,
+      "step": 3317
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3069629569107796e-05,
+      "loss": 0.223,
+      "step": 3318
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3065436220474822e-05,
+      "loss": 0.1968,
+      "step": 3319
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3061241984970055e-05,
+      "loss": 0.2129,
+      "step": 3320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.305704686305469e-05,
+      "loss": 0.2474,
+      "step": 3321
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3052850855190024e-05,
+      "loss": 0.2018,
+      "step": 3322
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3048653961837448e-05,
+      "loss": 0.2698,
+      "step": 3323
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3044456183458448e-05,
+      "loss": 0.2093,
+      "step": 3324
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3040257520514614e-05,
+      "loss": 0.2083,
+      "step": 3325
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3036057973467628e-05,
+      "loss": 0.1729,
+      "step": 3326
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3031857542779263e-05,
+      "loss": 0.1889,
+      "step": 3327
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3027656228911406e-05,
+      "loss": 0.2859,
+      "step": 3328
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3023454032326025e-05,
+      "loss": 0.2497,
+      "step": 3329
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3019250953485204e-05,
+      "loss": 0.293,
+      "step": 3330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3015046992851094e-05,
+      "loss": 0.132,
+      "step": 3331
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.301084215088597e-05,
+      "loss": 0.263,
+      "step": 3332
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3006636428052192e-05,
+      "loss": 0.2533,
+      "step": 3333
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.300242982481223e-05,
+      "loss": 0.2999,
+      "step": 3334
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.299822234162862e-05,
+      "loss": 0.1975,
+      "step": 3335
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2994013978964034e-05,
+      "loss": 0.2486,
+      "step": 3336
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2989804737281213e-05,
+      "loss": 0.2783,
+      "step": 3337
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2985594617043003e-05,
+      "loss": 0.122,
+      "step": 3338
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2981383618712343e-05,
+      "loss": 0.1971,
+      "step": 3339
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2977171742752287e-05,
+      "loss": 0.1589,
+      "step": 3340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2972958989625957e-05,
+      "loss": 0.2307,
+      "step": 3341
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2968745359796593e-05,
+      "loss": 0.2429,
+      "step": 3342
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.296453085372752e-05,
+      "loss": 0.1568,
+      "step": 3343
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2960315471882162e-05,
+      "loss": 0.1855,
+      "step": 3344
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2956099214724044e-05,
+      "loss": 0.2246,
+      "step": 3345
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2951882082716783e-05,
+      "loss": 0.2563,
+      "step": 3346
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2947664076324095e-05,
+      "loss": 0.2457,
+      "step": 3347
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2943445196009785e-05,
+      "loss": 0.1305,
+      "step": 3348
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2939225442237762e-05,
+      "loss": 0.2724,
+      "step": 3349
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2935004815472026e-05,
+      "loss": 0.2513,
+      "step": 3350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.293078331617669e-05,
+      "loss": 0.2106,
+      "step": 3351
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.292656094481592e-05,
+      "loss": 0.1746,
+      "step": 3352
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2922337701854028e-05,
+      "loss": 0.2191,
+      "step": 3353
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2918113587755395e-05,
+      "loss": 0.2316,
+      "step": 3354
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.29138886029845e-05,
+      "loss": 0.1913,
+      "step": 3355
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2909662748005923e-05,
+      "loss": 0.1659,
+      "step": 3356
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2905436023284334e-05,
+      "loss": 0.1685,
+      "step": 3357
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2901208429284502e-05,
+      "loss": 0.1961,
+      "step": 3358
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.28969799664713e-05,
+      "loss": 0.1826,
+      "step": 3359
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2892750635309675e-05,
+      "loss": 0.2128,
+      "step": 3360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.288852043626469e-05,
+      "loss": 0.1282,
+      "step": 3361
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2884289369801497e-05,
+      "loss": 0.2403,
+      "step": 3362
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2880057436385334e-05,
+      "loss": 0.1875,
+      "step": 3363
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.287582463648155e-05,
+      "loss": 0.2544,
+      "step": 3364
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2871590970555585e-05,
+      "loss": 0.209,
+      "step": 3365
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2867356439072956e-05,
+      "loss": 0.1468,
+      "step": 3366
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2863121042499308e-05,
+      "loss": 0.1928,
+      "step": 3367
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2858884781300353e-05,
+      "loss": 0.3094,
+      "step": 3368
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2854647655941913e-05,
+      "loss": 0.2246,
+      "step": 3369
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2850409666889892e-05,
+      "loss": 0.2612,
+      "step": 3370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2846170814610304e-05,
+      "loss": 0.196,
+      "step": 3371
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2841931099569255e-05,
+      "loss": 0.2073,
+      "step": 3372
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.283769052223294e-05,
+      "loss": 0.2604,
+      "step": 3373
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2833449083067645e-05,
+      "loss": 0.2338,
+      "step": 3374
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2829206782539755e-05,
+      "loss": 0.2049,
+      "step": 3375
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.282496362111577e-05,
+      "loss": 0.1788,
+      "step": 3376
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2820719599262243e-05,
+      "loss": 0.1642,
+      "step": 3377
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2816474717445857e-05,
+      "loss": 0.2011,
+      "step": 3378
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.281222897613338e-05,
+      "loss": 0.1017,
+      "step": 3379
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2807982375791666e-05,
+      "loss": 0.2678,
+      "step": 3380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2803734916887673e-05,
+      "loss": 0.2869,
+      "step": 3381
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2799486599888445e-05,
+      "loss": 0.2213,
+      "step": 3382
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2795237425261124e-05,
+      "loss": 0.1966,
+      "step": 3383
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2790987393472954e-05,
+      "loss": 0.2075,
+      "step": 3384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2786736504991263e-05,
+      "loss": 0.2076,
+      "step": 3385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.278248476028348e-05,
+      "loss": 0.1951,
+      "step": 3386
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2778232159817115e-05,
+      "loss": 0.2525,
+      "step": 3387
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2773978704059792e-05,
+      "loss": 0.257,
+      "step": 3388
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2769724393479217e-05,
+      "loss": 0.2414,
+      "step": 3389
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.27654692285432e-05,
+      "loss": 0.1935,
+      "step": 3390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2761213209719615e-05,
+      "loss": 0.1931,
+      "step": 3391
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2756956337476468e-05,
+      "loss": 0.2351,
+      "step": 3392
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2752698612281846e-05,
+      "loss": 0.2384,
+      "step": 3393
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.274844003460392e-05,
+      "loss": 0.3029,
+      "step": 3394
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.274418060491096e-05,
+      "loss": 0.2825,
+      "step": 3395
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2739920323671334e-05,
+      "loss": 0.1585,
+      "step": 3396
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.27356591913535e-05,
+      "loss": 0.168,
+      "step": 3397
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2731397208426005e-05,
+      "loss": 0.2423,
+      "step": 3398
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2727134375357505e-05,
+      "loss": 0.2441,
+      "step": 3399
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2722870692616727e-05,
+      "loss": 0.2765,
+      "step": 3400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.271860616067252e-05,
+      "loss": 0.2087,
+      "step": 3401
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2714340779993793e-05,
+      "loss": 0.2173,
+      "step": 3402
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2710074551049576e-05,
+      "loss": 0.1213,
+      "step": 3403
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.270580747430898e-05,
+      "loss": 0.2977,
+      "step": 3404
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.270153955024121e-05,
+      "loss": 0.228,
+      "step": 3405
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2697270779315565e-05,
+      "loss": 0.2522,
+      "step": 3406
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2693001162001435e-05,
+      "loss": 0.2861,
+      "step": 3407
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2688730698768304e-05,
+      "loss": 0.1759,
+      "step": 3408
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2684459390085758e-05,
+      "loss": 0.3213,
+      "step": 3409
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.268018723642346e-05,
+      "loss": 0.2223,
+      "step": 3410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2675914238251184e-05,
+      "loss": 0.3145,
+      "step": 3411
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2671640396038773e-05,
+      "loss": 0.2436,
+      "step": 3412
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2667365710256188e-05,
+      "loss": 0.2515,
+      "step": 3413
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2663090181373463e-05,
+      "loss": 0.2114,
+      "step": 3414
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2658813809860747e-05,
+      "loss": 0.2866,
+      "step": 3415
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2654536596188253e-05,
+      "loss": 0.1332,
+      "step": 3416
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.265025854082631e-05,
+      "loss": 0.2455,
+      "step": 3417
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2645979644245327e-05,
+      "loss": 0.2519,
+      "step": 3418
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.264169990691581e-05,
+      "loss": 0.1158,
+      "step": 3419
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.263741932930836e-05,
+      "loss": 0.2375,
+      "step": 3420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2633137911893665e-05,
+      "loss": 0.2372,
+      "step": 3421
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2628855655142505e-05,
+      "loss": 0.1849,
+      "step": 3422
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2624572559525764e-05,
+      "loss": 0.2009,
+      "step": 3423
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2620288625514402e-05,
+      "loss": 0.2182,
+      "step": 3424
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2616003853579482e-05,
+      "loss": 0.1798,
+      "step": 3425
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.261171824419215e-05,
+      "loss": 0.2296,
+      "step": 3426
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2607431797823658e-05,
+      "loss": 0.256,
+      "step": 3427
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2603144514945333e-05,
+      "loss": 0.2229,
+      "step": 3428
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2598856396028612e-05,
+      "loss": 0.1963,
+      "step": 3429
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2594567441545003e-05,
+      "loss": 0.217,
+      "step": 3430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2590277651966128e-05,
+      "loss": 0.1417,
+      "step": 3431
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2585987027763685e-05,
+      "loss": 0.1818,
+      "step": 3432
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2581695569409473e-05,
+      "loss": 0.2909,
+      "step": 3433
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.257740327737538e-05,
+      "loss": 0.224,
+      "step": 3434
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2573110152133375e-05,
+      "loss": 0.2383,
+      "step": 3435
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2568816194155534e-05,
+      "loss": 0.2336,
+      "step": 3436
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2564521403914022e-05,
+      "loss": 0.1926,
+      "step": 3437
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.256022578188109e-05,
+      "loss": 0.2134,
+      "step": 3438
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2555929328529075e-05,
+      "loss": 0.1638,
+      "step": 3439
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.255163204433043e-05,
+      "loss": 0.2212,
+      "step": 3440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2547333929757665e-05,
+      "loss": 0.2257,
+      "step": 3441
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2543034985283406e-05,
+      "loss": 0.2295,
+      "step": 3442
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.253873521138037e-05,
+      "loss": 0.2736,
+      "step": 3443
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.253443460852135e-05,
+      "loss": 0.1242,
+      "step": 3444
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2530133177179235e-05,
+      "loss": 0.2342,
+      "step": 3445
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2525830917827015e-05,
+      "loss": 0.2618,
+      "step": 3446
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.252152783093776e-05,
+      "loss": 0.1719,
+      "step": 3447
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2517223916984646e-05,
+      "loss": 0.2739,
+      "step": 3448
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2512919176440918e-05,
+      "loss": 0.3134,
+      "step": 3449
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2508613609779926e-05,
+      "loss": 0.186,
+      "step": 3450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2504307217475113e-05,
+      "loss": 0.3644,
+      "step": 3451
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.25e-05,
+      "loss": 0.1919,
+      "step": 3452
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2495691957828215e-05,
+      "loss": 0.1799,
+      "step": 3453
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2491383091433462e-05,
+      "loss": 0.1811,
+      "step": 3454
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2487073401289546e-05,
+      "loss": 0.2196,
+      "step": 3455
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2482762887870358e-05,
+      "loss": 0.2099,
+      "step": 3456
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.247845155164988e-05,
+      "loss": 0.1897,
+      "step": 3457
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2474139393102182e-05,
+      "loss": 0.2441,
+      "step": 3458
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.246982641270143e-05,
+      "loss": 0.1966,
+      "step": 3459
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2465512610921876e-05,
+      "loss": 0.2462,
+      "step": 3460
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.246119798823786e-05,
+      "loss": 0.2502,
+      "step": 3461
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2456882545123824e-05,
+      "loss": 0.2017,
+      "step": 3462
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2452566282054287e-05,
+      "loss": 0.1768,
+      "step": 3463
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2448249199503866e-05,
+      "loss": 0.1457,
+      "step": 3464
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2443931297947266e-05,
+      "loss": 0.1812,
+      "step": 3465
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2439612577859277e-05,
+      "loss": 0.1586,
+      "step": 3466
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.243529303971479e-05,
+      "loss": 0.1771,
+      "step": 3467
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2430972683988774e-05,
+      "loss": 0.2337,
+      "step": 3468
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2426651511156295e-05,
+      "loss": 0.2257,
+      "step": 3469
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2422329521692508e-05,
+      "loss": 0.1952,
+      "step": 3470
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.241800671607266e-05,
+      "loss": 0.1701,
+      "step": 3471
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.241368309477207e-05,
+      "loss": 0.1962,
+      "step": 3472
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2409358658266184e-05,
+      "loss": 0.2163,
+      "step": 3473
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.24050334070305e-05,
+      "loss": 0.2559,
+      "step": 3474
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2400707341540625e-05,
+      "loss": 0.1963,
+      "step": 3475
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.239638046227225e-05,
+      "loss": 0.1743,
+      "step": 3476
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2392052769701157e-05,
+      "loss": 0.2086,
+      "step": 3477
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.238772426430322e-05,
+      "loss": 0.197,
+      "step": 3478
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.23833949465544e-05,
+      "loss": 0.2177,
+      "step": 3479
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2379064816930737e-05,
+      "loss": 0.206,
+      "step": 3480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.237473387590838e-05,
+      "loss": 0.2047,
+      "step": 3481
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2370402123963557e-05,
+      "loss": 0.2334,
+      "step": 3482
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.236606956157258e-05,
+      "loss": 0.2162,
+      "step": 3483
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.236173618921186e-05,
+      "loss": 0.2085,
+      "step": 3484
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2357402007357893e-05,
+      "loss": 0.1757,
+      "step": 3485
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.235306701648726e-05,
+      "loss": 0.2242,
+      "step": 3486
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.234873121707664e-05,
+      "loss": 0.1707,
+      "step": 3487
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.234439460960279e-05,
+      "loss": 0.2377,
+      "step": 3488
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2340057194542565e-05,
+      "loss": 0.2064,
+      "step": 3489
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.23357189723729e-05,
+      "loss": 0.2119,
+      "step": 3490
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2331379943570837e-05,
+      "loss": 0.2171,
+      "step": 3491
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2327040108613477e-05,
+      "loss": 0.2368,
+      "step": 3492
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.232269946797804e-05,
+      "loss": 0.22,
+      "step": 3493
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2318358022141812e-05,
+      "loss": 0.1202,
+      "step": 3494
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2314015771582177e-05,
+      "loss": 0.2904,
+      "step": 3495
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.230967271677661e-05,
+      "loss": 0.246,
+      "step": 3496
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2305328858202674e-05,
+      "loss": 0.1602,
+      "step": 3497
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2300984196338017e-05,
+      "loss": 0.2175,
+      "step": 3498
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2296638731660368e-05,
+      "loss": 0.1845,
+      "step": 3499
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.229229246464756e-05,
+      "loss": 0.2677,
+      "step": 3500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2287945395777507e-05,
+      "loss": 0.2604,
+      "step": 3501
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.228359752552821e-05,
+      "loss": 0.2062,
+      "step": 3502
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.227924885437775e-05,
+      "loss": 0.2745,
+      "step": 3503
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2274899382804316e-05,
+      "loss": 0.2244,
+      "step": 3504
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.227054911128617e-05,
+      "loss": 0.2157,
+      "step": 3505
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2266198040301667e-05,
+      "loss": 0.2409,
+      "step": 3506
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.226184617032925e-05,
+      "loss": 0.1967,
+      "step": 3507
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2257493501847444e-05,
+      "loss": 0.2435,
+      "step": 3508
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.225314003533487e-05,
+      "loss": 0.179,
+      "step": 3509
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2248785771270233e-05,
+      "loss": 0.1789,
+      "step": 3510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2244430710132323e-05,
+      "loss": 0.2297,
+      "step": 3511
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.224007485240003e-05,
+      "loss": 0.1213,
+      "step": 3512
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2235718198552306e-05,
+      "loss": 0.1843,
+      "step": 3513
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2231360749068223e-05,
+      "loss": 0.2111,
+      "step": 3514
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.222700250442691e-05,
+      "loss": 0.2321,
+      "step": 3515
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.222264346510761e-05,
+      "loss": 0.2077,
+      "step": 3516
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2218283631589633e-05,
+      "loss": 0.274,
+      "step": 3517
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.221392300435239e-05,
+      "loss": 0.168,
+      "step": 3518
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2209561583875367e-05,
+      "loss": 0.3002,
+      "step": 3519
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.220519937063815e-05,
+      "loss": 0.1978,
+      "step": 3520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2200836365120407e-05,
+      "loss": 0.2333,
+      "step": 3521
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.219647256780189e-05,
+      "loss": 0.3143,
+      "step": 3522
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2192107979162433e-05,
+      "loss": 0.1808,
+      "step": 3523
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.218774259968197e-05,
+      "loss": 0.1775,
+      "step": 3524
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.218337642984052e-05,
+      "loss": 0.152,
+      "step": 3525
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.217900947011818e-05,
+      "loss": 0.205,
+      "step": 3526
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.217464172099514e-05,
+      "loss": 0.3266,
+      "step": 3527
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2170273182951676e-05,
+      "loss": 0.189,
+      "step": 3528
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2165903856468155e-05,
+      "loss": 0.265,
+      "step": 3529
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2161533742025017e-05,
+      "loss": 0.1449,
+      "step": 3530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2157162840102806e-05,
+      "loss": 0.182,
+      "step": 3531
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.215279115118214e-05,
+      "loss": 0.2091,
+      "step": 3532
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2148418675743734e-05,
+      "loss": 0.2083,
+      "step": 3533
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.214404541426837e-05,
+      "loss": 0.1748,
+      "step": 3534
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2139671367236946e-05,
+      "loss": 0.1812,
+      "step": 3535
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.213529653513042e-05,
+      "loss": 0.2529,
+      "step": 3536
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2130920918429852e-05,
+      "loss": 0.1832,
+      "step": 3537
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2126544517616375e-05,
+      "loss": 0.2202,
+      "step": 3538
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2122167333171224e-05,
+      "loss": 0.1344,
+      "step": 3539
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2117789365575714e-05,
+      "loss": 0.2362,
+      "step": 3540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2113410615311236e-05,
+      "loss": 0.1459,
+      "step": 3541
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2109031082859276e-05,
+      "loss": 0.1431,
+      "step": 3542
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.210465076870142e-05,
+      "loss": 0.1315,
+      "step": 3543
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2100269673319305e-05,
+      "loss": 0.1673,
+      "step": 3544
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2095887797194685e-05,
+      "loss": 0.1611,
+      "step": 3545
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.209150514080939e-05,
+      "loss": 0.2785,
+      "step": 3546
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.208712170464533e-05,
+      "loss": 0.2073,
+      "step": 3547
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2082737489184515e-05,
+      "loss": 0.1834,
+      "step": 3548
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.207835249490902e-05,
+      "loss": 0.2193,
+      "step": 3549
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2073966722301016e-05,
+      "loss": 0.1333,
+      "step": 3550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.206958017184278e-05,
+      "loss": 0.2954,
+      "step": 3551
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.206519284401663e-05,
+      "loss": 0.1934,
+      "step": 3552
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2060804739305008e-05,
+      "loss": 0.205,
+      "step": 3553
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2056415858190437e-05,
+      "loss": 0.1851,
+      "step": 3554
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2052026201155495e-05,
+      "loss": 0.1755,
+      "step": 3555
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.204763576868288e-05,
+      "loss": 0.2151,
+      "step": 3556
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.204324456125536e-05,
+      "loss": 0.2555,
+      "step": 3557
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2038852579355793e-05,
+      "loss": 0.1539,
+      "step": 3558
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.203445982346711e-05,
+      "loss": 0.1631,
+      "step": 3559
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.203006629407235e-05,
+      "loss": 0.1838,
+      "step": 3560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.202567199165461e-05,
+      "loss": 0.1552,
+      "step": 3561
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2021276916697097e-05,
+      "loss": 0.1861,
+      "step": 3562
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.201688106968308e-05,
+      "loss": 0.1784,
+      "step": 3563
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2012484451095933e-05,
+      "loss": 0.2626,
+      "step": 3564
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2008087061419106e-05,
+      "loss": 0.2288,
+      "step": 3565
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.200368890113613e-05,
+      "loss": 0.2861,
+      "step": 3566
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1999289970730625e-05,
+      "loss": 0.2327,
+      "step": 3567
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.19948902706863e-05,
+      "loss": 0.1736,
+      "step": 3568
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.199048980148694e-05,
+      "loss": 0.2678,
+      "step": 3569
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.198608856361642e-05,
+      "loss": 0.1644,
+      "step": 3570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.19816865575587e-05,
+      "loss": 0.177,
+      "step": 3571
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.197728378379782e-05,
+      "loss": 0.1717,
+      "step": 3572
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1972880242817908e-05,
+      "loss": 0.155,
+      "step": 3573
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.196847593510317e-05,
+      "loss": 0.1672,
+      "step": 3574
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.196407086113791e-05,
+      "loss": 0.2438,
+      "step": 3575
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1959665021406508e-05,
+      "loss": 0.242,
+      "step": 3576
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.195525841639342e-05,
+      "loss": 0.2566,
+      "step": 3577
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.19508510465832e-05,
+      "loss": 0.3156,
+      "step": 3578
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.194644291246048e-05,
+      "loss": 0.1432,
+      "step": 3579
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1942034014509977e-05,
+      "loss": 0.1609,
+      "step": 3580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.193762435321649e-05,
+      "loss": 0.1753,
+      "step": 3581
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1933213929064904e-05,
+      "loss": 0.15,
+      "step": 3582
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.192880274254019e-05,
+      "loss": 0.1474,
+      "step": 3583
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1924390794127397e-05,
+      "loss": 0.3427,
+      "step": 3584
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.191997808431166e-05,
+      "loss": 0.2718,
+      "step": 3585
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1915564613578198e-05,
+      "loss": 0.1941,
+      "step": 3586
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1911150382412317e-05,
+      "loss": 0.213,
+      "step": 3587
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1906735391299403e-05,
+      "loss": 0.2185,
+      "step": 3588
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1902319640724928e-05,
+      "loss": 0.2848,
+      "step": 3589
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1897903131174447e-05,
+      "loss": 0.235,
+      "step": 3590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.189348586313359e-05,
+      "loss": 0.2825,
+      "step": 3591
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1889067837088082e-05,
+      "loss": 0.3181,
+      "step": 3592
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1884649053523735e-05,
+      "loss": 0.1673,
+      "step": 3593
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1880229512926426e-05,
+      "loss": 0.1452,
+      "step": 3594
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1875809215782127e-05,
+      "loss": 0.1662,
+      "step": 3595
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1871388162576898e-05,
+      "loss": 0.2366,
+      "step": 3596
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.186696635379687e-05,
+      "loss": 0.1593,
+      "step": 3597
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1862543789928263e-05,
+      "loss": 0.2236,
+      "step": 3598
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1858120471457388e-05,
+      "loss": 0.203,
+      "step": 3599
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1853696398870623e-05,
+      "loss": 0.1938,
+      "step": 3600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.184927157265444e-05,
+      "loss": 0.2542,
+      "step": 3601
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1844845993295386e-05,
+      "loss": 0.2486,
+      "step": 3602
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1840419661280106e-05,
+      "loss": 0.2499,
+      "step": 3603
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1835992577095314e-05,
+      "loss": 0.2835,
+      "step": 3604
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1831564741227803e-05,
+      "loss": 0.231,
+      "step": 3605
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1827136154164457e-05,
+      "loss": 0.1596,
+      "step": 3606
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1822706816392255e-05,
+      "loss": 0.2488,
+      "step": 3607
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1818276728398233e-05,
+      "loss": 0.2212,
+      "step": 3608
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.181384589066952e-05,
+      "loss": 0.2476,
+      "step": 3609
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1809414303693336e-05,
+      "loss": 0.1565,
+      "step": 3610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.180498196795697e-05,
+      "loss": 0.2588,
+      "step": 3611
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.180054888394781e-05,
+      "loss": 0.2156,
+      "step": 3612
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.17961150521533e-05,
+      "loss": 0.1689,
+      "step": 3613
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1791680473060998e-05,
+      "loss": 0.2626,
+      "step": 3614
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.178724514715852e-05,
+      "loss": 0.2752,
+      "step": 3615
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1782809074933568e-05,
+      "loss": 0.2172,
+      "step": 3616
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.177837225687394e-05,
+      "loss": 0.1255,
+      "step": 3617
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.177393469346751e-05,
+      "loss": 0.1577,
+      "step": 3618
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1769496385202213e-05,
+      "loss": 0.1058,
+      "step": 3619
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.17650573325661e-05,
+      "loss": 0.202,
+      "step": 3620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.176061753604728e-05,
+      "loss": 0.2707,
+      "step": 3621
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1756176996133952e-05,
+      "loss": 0.2379,
+      "step": 3622
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1751735713314396e-05,
+      "loss": 0.1964,
+      "step": 3623
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1747293688076974e-05,
+      "loss": 0.1703,
+      "step": 3624
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1742850920910126e-05,
+      "loss": 0.2013,
+      "step": 3625
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1738407412302382e-05,
+      "loss": 0.1746,
+      "step": 3626
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1733963162742345e-05,
+      "loss": 0.2121,
+      "step": 3627
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1729518172718704e-05,
+      "loss": 0.1483,
+      "step": 3628
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1725072442720225e-05,
+      "loss": 0.1919,
+      "step": 3629
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1720625973235762e-05,
+      "loss": 0.2025,
+      "step": 3630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1716178764754244e-05,
+      "loss": 0.2394,
+      "step": 3631
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1711730817764692e-05,
+      "loss": 0.2043,
+      "step": 3632
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1707282132756184e-05,
+      "loss": 0.1631,
+      "step": 3633
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1702832710217915e-05,
+      "loss": 0.2398,
+      "step": 3634
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1698382550639127e-05,
+      "loss": 0.198,
+      "step": 3635
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1693931654509166e-05,
+      "loss": 0.2207,
+      "step": 3636
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1689480022317446e-05,
+      "loss": 0.2111,
+      "step": 3637
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.168502765455347e-05,
+      "loss": 0.1621,
+      "step": 3638
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.168057455170681e-05,
+      "loss": 0.1535,
+      "step": 3639
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.167612071426714e-05,
+      "loss": 0.2297,
+      "step": 3640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1671666142724186e-05,
+      "loss": 0.1401,
+      "step": 3641
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1667210837567787e-05,
+      "loss": 0.152,
+      "step": 3642
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1662754799287837e-05,
+      "loss": 0.2966,
+      "step": 3643
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1658298028374323e-05,
+      "loss": 0.2366,
+      "step": 3644
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1653840525317307e-05,
+      "loss": 0.2413,
+      "step": 3645
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1649382290606942e-05,
+      "loss": 0.2396,
+      "step": 3646
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.164492332473344e-05,
+      "loss": 0.1595,
+      "step": 3647
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1640463628187124e-05,
+      "loss": 0.2184,
+      "step": 3648
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1636003201458363e-05,
+      "loss": 0.2432,
+      "step": 3649
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1631542045037632e-05,
+      "loss": 0.2271,
+      "step": 3650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.162708015941548e-05,
+      "loss": 0.241,
+      "step": 3651
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.162261754508253e-05,
+      "loss": 0.1964,
+      "step": 3652
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1618154202529488e-05,
+      "loss": 0.183,
+      "step": 3653
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.161369013224714e-05,
+      "loss": 0.2368,
+      "step": 3654
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1609225334726365e-05,
+      "loss": 0.1622,
+      "step": 3655
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1604759810458093e-05,
+      "loss": 0.2103,
+      "step": 3656
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1600293559933367e-05,
+      "loss": 0.2602,
+      "step": 3657
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1595826583643284e-05,
+      "loss": 0.2606,
+      "step": 3658
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.159135888207903e-05,
+      "loss": 0.1872,
+      "step": 3659
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.158689045573188e-05,
+      "loss": 0.1589,
+      "step": 3660
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.158242130509317e-05,
+      "loss": 0.1969,
+      "step": 3661
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1577951430654334e-05,
+      "loss": 0.1898,
+      "step": 3662
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1573480832906872e-05,
+      "loss": 0.1767,
+      "step": 3663
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.156900951234237e-05,
+      "loss": 0.1735,
+      "step": 3664
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.15645374694525e-05,
+      "loss": 0.261,
+      "step": 3665
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1560064704728995e-05,
+      "loss": 0.2059,
+      "step": 3666
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1555591218663683e-05,
+      "loss": 0.1566,
+      "step": 3667
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1551117011748466e-05,
+      "loss": 0.2744,
+      "step": 3668
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1546642084475327e-05,
+      "loss": 0.2822,
+      "step": 3669
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1542166437336323e-05,
+      "loss": 0.2084,
+      "step": 3670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1537690070823604e-05,
+      "loss": 0.245,
+      "step": 3671
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1533212985429374e-05,
+      "loss": 0.1383,
+      "step": 3672
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1528735181645946e-05,
+      "loss": 0.1904,
+      "step": 3673
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.152425665996569e-05,
+      "loss": 0.2513,
+      "step": 3674
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.151977742088106e-05,
+      "loss": 0.2534,
+      "step": 3675
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.15152974648846e-05,
+      "loss": 0.2531,
+      "step": 3676
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.151081679246891e-05,
+      "loss": 0.2194,
+      "step": 3677
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.15063354041267e-05,
+      "loss": 0.2489,
+      "step": 3678
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.150185330035073e-05,
+      "loss": 0.1847,
+      "step": 3679
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.149737048163385e-05,
+      "loss": 0.1659,
+      "step": 3680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1492886948468992e-05,
+      "loss": 0.192,
+      "step": 3681
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.148840270134917e-05,
+      "loss": 0.171,
+      "step": 3682
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1483917740767452e-05,
+      "loss": 0.2094,
+      "step": 3683
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1479432067217022e-05,
+      "loss": 0.1793,
+      "step": 3684
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.147494568119111e-05,
+      "loss": 0.1904,
+      "step": 3685
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.147045858318304e-05,
+      "loss": 0.2388,
+      "step": 3686
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.146597077368621e-05,
+      "loss": 0.1807,
+      "step": 3687
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1461482253194108e-05,
+      "loss": 0.2022,
+      "step": 3688
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.145699302220027e-05,
+      "loss": 0.2958,
+      "step": 3689
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1452503081198354e-05,
+      "loss": 0.2103,
+      "step": 3690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1448012430682045e-05,
+      "loss": 0.244,
+      "step": 3691
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.144352107114516e-05,
+      "loss": 0.1719,
+      "step": 3692
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.143902900308155e-05,
+      "loss": 0.1773,
+      "step": 3693
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.143453622698516e-05,
+      "loss": 0.1267,
+      "step": 3694
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1430042743350017e-05,
+      "loss": 0.1975,
+      "step": 3695
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1425548552670233e-05,
+      "loss": 0.1543,
+      "step": 3696
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.142105365543997e-05,
+      "loss": 0.2153,
+      "step": 3697
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.14165580521535e-05,
+      "loss": 0.2643,
+      "step": 3698
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1412061743305145e-05,
+      "loss": 0.1723,
+      "step": 3699
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1407564729389327e-05,
+      "loss": 0.2346,
+      "step": 3700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.140306701090053e-05,
+      "loss": 0.2734,
+      "step": 3701
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.139856858833332e-05,
+      "loss": 0.2039,
+      "step": 3702
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1394069462182343e-05,
+      "loss": 0.1393,
+      "step": 3703
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1389569632942325e-05,
+      "loss": 0.1722,
+      "step": 3704
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.138506910110806e-05,
+      "loss": 0.2833,
+      "step": 3705
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1380567867174428e-05,
+      "loss": 0.1612,
+      "step": 3706
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1376065931636385e-05,
+      "loss": 0.227,
+      "step": 3707
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.137156329498895e-05,
+      "loss": 0.1967,
+      "step": 3708
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.136705995772725e-05,
+      "loss": 0.1561,
+      "step": 3709
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.136255592034645e-05,
+      "loss": 0.1866,
+      "step": 3710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1358051183341824e-05,
+      "loss": 0.221,
+      "step": 3711
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1353545747208713e-05,
+      "loss": 0.167,
+      "step": 3712
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.134903961244252e-05,
+      "loss": 0.2414,
+      "step": 3713
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1344532779538746e-05,
+      "loss": 0.1537,
+      "step": 3714
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1340025248992966e-05,
+      "loss": 0.2431,
+      "step": 3715
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.133551702130081e-05,
+      "loss": 0.116,
+      "step": 3716
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1331008096958016e-05,
+      "loss": 0.1849,
+      "step": 3717
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1326498476460377e-05,
+      "loss": 0.1621,
+      "step": 3718
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1321988160303774e-05,
+      "loss": 0.2442,
+      "step": 3719
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1317477148984148e-05,
+      "loss": 0.1814,
+      "step": 3720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.131296544299754e-05,
+      "loss": 0.1857,
+      "step": 3721
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1308453042840046e-05,
+      "loss": 0.2809,
+      "step": 3722
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1303939949007855e-05,
+      "loss": 0.2327,
+      "step": 3723
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.129942616199722e-05,
+      "loss": 0.2296,
+      "step": 3724
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1294911682304472e-05,
+      "loss": 0.2208,
+      "step": 3725
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.129039651042603e-05,
+      "loss": 0.2872,
+      "step": 3726
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1285880646858375e-05,
+      "loss": 0.098,
+      "step": 3727
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1281364092098066e-05,
+      "loss": 0.2012,
+      "step": 3728
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1276846846641752e-05,
+      "loss": 0.2451,
+      "step": 3729
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1272328910986137e-05,
+      "loss": 0.189,
+      "step": 3730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1267810285628018e-05,
+      "loss": 0.2094,
+      "step": 3731
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1263290971064256e-05,
+      "loss": 0.1558,
+      "step": 3732
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1258770967791794e-05,
+      "loss": 0.1512,
+      "step": 3733
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.125425027630765e-05,
+      "loss": 0.2407,
+      "step": 3734
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1249728897108923e-05,
+      "loss": 0.361,
+      "step": 3735
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.124520683069277e-05,
+      "loss": 0.2617,
+      "step": 3736
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.124068407755645e-05,
+      "loss": 0.1499,
+      "step": 3737
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1236160638197264e-05,
+      "loss": 0.1786,
+      "step": 3738
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.123163651311262e-05,
+      "loss": 0.1492,
+      "step": 3739
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1227111702799996e-05,
+      "loss": 0.1864,
+      "step": 3740
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1222586207756922e-05,
+      "loss": 0.1857,
+      "step": 3741
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.121806002848103e-05,
+      "loss": 0.2183,
+      "step": 3742
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.121353316547001e-05,
+      "loss": 0.1692,
+      "step": 3743
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.120900561922164e-05,
+      "loss": 0.2148,
+      "step": 3744
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.120447739023376e-05,
+      "loss": 0.2162,
+      "step": 3745
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1199948479004306e-05,
+      "loss": 0.2645,
+      "step": 3746
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1195418886031252e-05,
+      "loss": 0.1876,
+      "step": 3747
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1190888611812692e-05,
+      "loss": 0.1545,
+      "step": 3748
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1186357656846763e-05,
+      "loss": 0.174,
+      "step": 3749
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1181826021631685e-05,
+      "loss": 0.2655,
+      "step": 3750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.117729370666576e-05,
+      "loss": 0.2201,
+      "step": 3751
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.117276071244736e-05,
+      "loss": 0.2065,
+      "step": 3752
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.116822703947492e-05,
+      "loss": 0.1384,
+      "step": 3753
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1163692688246975e-05,
+      "loss": 0.2004,
+      "step": 3754
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.115915765926211e-05,
+      "loss": 0.1894,
+      "step": 3755
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1154621953018998e-05,
+      "loss": 0.1763,
+      "step": 3756
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.115008557001639e-05,
+      "loss": 0.1656,
+      "step": 3757
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1145548510753096e-05,
+      "loss": 0.2056,
+      "step": 3758
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1141010775728007e-05,
+      "loss": 0.2437,
+      "step": 3759
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.11364723654401e-05,
+      "loss": 0.1806,
+      "step": 3760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1131933280388407e-05,
+      "loss": 0.2546,
+      "step": 3761
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1127393521072055e-05,
+      "loss": 0.148,
+      "step": 3762
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1122853087990222e-05,
+      "loss": 0.2239,
+      "step": 3763
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1118311981642177e-05,
+      "loss": 0.1961,
+      "step": 3764
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1113770202527264e-05,
+      "loss": 0.1574,
+      "step": 3765
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1109227751144882e-05,
+      "loss": 0.1997,
+      "step": 3766
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1104684627994528e-05,
+      "loss": 0.1483,
+      "step": 3767
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1100140833575756e-05,
+      "loss": 0.13,
+      "step": 3768
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1095596368388206e-05,
+      "loss": 0.2018,
+      "step": 3769
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1091051232931575e-05,
+      "loss": 0.2523,
+      "step": 3770
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1086505427705657e-05,
+      "loss": 0.2153,
+      "step": 3771
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1081958953210297e-05,
+      "loss": 0.363,
+      "step": 3772
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1077411809945423e-05,
+      "loss": 0.207,
+      "step": 3773
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1072863998411045e-05,
+      "loss": 0.1127,
+      "step": 3774
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.106831551910723e-05,
+      "loss": 0.2191,
+      "step": 3775
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1063766372534133e-05,
+      "loss": 0.1949,
+      "step": 3776
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1059216559191972e-05,
+      "loss": 0.2479,
+      "step": 3777
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1054666079581036e-05,
+      "loss": 0.2207,
+      "step": 3778
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1050114934201716e-05,
+      "loss": 0.2509,
+      "step": 3779
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.104556312355443e-05,
+      "loss": 0.2023,
+      "step": 3780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1041010648139706e-05,
+      "loss": 0.2115,
+      "step": 3781
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1036457508458127e-05,
+      "loss": 0.1943,
+      "step": 3782
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1031903705010353e-05,
+      "loss": 0.1654,
+      "step": 3783
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1027349238297126e-05,
+      "loss": 0.2036,
+      "step": 3784
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1022794108819246e-05,
+      "loss": 0.2415,
+      "step": 3785
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1018238317077594e-05,
+      "loss": 0.192,
+      "step": 3786
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1013681863573125e-05,
+      "loss": 0.1973,
+      "step": 3787
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1009124748806867e-05,
+      "loss": 0.246,
+      "step": 3788
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.100456697327991e-05,
+      "loss": 0.2174,
+      "step": 3789
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1000008537493434e-05,
+      "loss": 0.2081,
+      "step": 3790
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0995449441948676e-05,
+      "loss": 0.1866,
+      "step": 3791
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0990889687146953e-05,
+      "loss": 0.2047,
+      "step": 3792
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0986329273589662e-05,
+      "loss": 0.2335,
+      "step": 3793
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.098176820177825e-05,
+      "loss": 0.2327,
+      "step": 3794
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.097720647221426e-05,
+      "loss": 0.1537,
+      "step": 3795
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.09726440853993e-05,
+      "loss": 0.1804,
+      "step": 3796
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.096808104183504e-05,
+      "loss": 0.1768,
+      "step": 3797
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.096351734202323e-05,
+      "loss": 0.162,
+      "step": 3798
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0958952986465703e-05,
+      "loss": 0.3018,
+      "step": 3799
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0954387975664345e-05,
+      "loss": 0.2587,
+      "step": 3800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.094982231012112e-05,
+      "loss": 0.1516,
+      "step": 3801
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0945255990338075e-05,
+      "loss": 0.2028,
+      "step": 3802
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0940689016817315e-05,
+      "loss": 0.2228,
+      "step": 3803
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.093612139006103e-05,
+      "loss": 0.1139,
+      "step": 3804
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0931553110571457e-05,
+      "loss": 0.1632,
+      "step": 3805
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.092698417885094e-05,
+      "loss": 0.2344,
+      "step": 3806
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0922414595401867e-05,
+      "loss": 0.1952,
+      "step": 3807
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0917844360726713e-05,
+      "loss": 0.153,
+      "step": 3808
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0913273475328013e-05,
+      "loss": 0.2033,
+      "step": 3809
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0908701939708385e-05,
+      "loss": 0.1715,
+      "step": 3810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0904129754370505e-05,
+      "loss": 0.1989,
+      "step": 3811
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.089955691981714e-05,
+      "loss": 0.2368,
+      "step": 3812
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0894983436551107e-05,
+      "loss": 0.2582,
+      "step": 3813
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0890409305075315e-05,
+      "loss": 0.1894,
+      "step": 3814
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0885834525892723e-05,
+      "loss": 0.1694,
+      "step": 3815
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0881259099506372e-05,
+      "loss": 0.1892,
+      "step": 3816
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0876683026419377e-05,
+      "loss": 0.1956,
+      "step": 3817
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.087210630713493e-05,
+      "loss": 0.2436,
+      "step": 3818
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.086752894215626e-05,
+      "loss": 0.2163,
+      "step": 3819
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0862950931986723e-05,
+      "loss": 0.1521,
+      "step": 3820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0858372277129695e-05,
+      "loss": 0.1908,
+      "step": 3821
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0853792978088648e-05,
+      "loss": 0.1823,
+      "step": 3822
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0849213035367118e-05,
+      "loss": 0.1335,
+      "step": 3823
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.084463244946872e-05,
+      "loss": 0.199,
+      "step": 3824
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.084005122089712e-05,
+      "loss": 0.1212,
+      "step": 3825
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.083546935015608e-05,
+      "loss": 0.3032,
+      "step": 3826
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.083088683774942e-05,
+      "loss": 0.2727,
+      "step": 3827
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0826303684181023e-05,
+      "loss": 0.1629,
+      "step": 3828
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.082171988995486e-05,
+      "loss": 0.1295,
+      "step": 3829
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0817135455574956e-05,
+      "loss": 0.2245,
+      "step": 3830
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0812550381545416e-05,
+      "loss": 0.1652,
+      "step": 3831
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.080796466837041e-05,
+      "loss": 0.1244,
+      "step": 3832
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0803378316554184e-05,
+      "loss": 0.215,
+      "step": 3833
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.079879132660105e-05,
+      "loss": 0.1537,
+      "step": 3834
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0794203699015396e-05,
+      "loss": 0.1762,
+      "step": 3835
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.078961543430167e-05,
+      "loss": 0.2634,
+      "step": 3836
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0785026532964395e-05,
+      "loss": 0.2322,
+      "step": 3837
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.078043699550817e-05,
+      "loss": 0.1797,
+      "step": 3838
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.077584682243765e-05,
+      "loss": 0.224,
+      "step": 3839
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.077125601425758e-05,
+      "loss": 0.2023,
+      "step": 3840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.076666457147275e-05,
+      "loss": 0.2174,
+      "step": 3841
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.076207249458804e-05,
+      "loss": 0.1667,
+      "step": 3842
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.07574797841084e-05,
+      "loss": 0.2053,
+      "step": 3843
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0752886440538825e-05,
+      "loss": 0.1832,
+      "step": 3844
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.074829246438441e-05,
+      "loss": 0.2061,
+      "step": 3845
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0743697856150306e-05,
+      "loss": 0.1263,
+      "step": 3846
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0739102616341733e-05,
+      "loss": 0.1943,
+      "step": 3847
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0734506745463972e-05,
+      "loss": 0.2427,
+      "step": 3848
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.07299102440224e-05,
+      "loss": 0.2338,
+      "step": 3849
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.072531311252243e-05,
+      "loss": 0.2036,
+      "step": 3850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.072071535146957e-05,
+      "loss": 0.2611,
+      "step": 3851
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0716116961369388e-05,
+      "loss": 0.1613,
+      "step": 3852
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0711517942727517e-05,
+      "loss": 0.2431,
+      "step": 3853
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0706918296049666e-05,
+      "loss": 0.1839,
+      "step": 3854
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0702318021841604e-05,
+      "loss": 0.1276,
+      "step": 3855
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.069771712060918e-05,
+      "loss": 0.1182,
+      "step": 3856
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.069311559285831e-05,
+      "loss": 0.1656,
+      "step": 3857
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0688513439094974e-05,
+      "loss": 0.1926,
+      "step": 3858
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0683910659825216e-05,
+      "loss": 0.2411,
+      "step": 3859
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.067930725555517e-05,
+      "loss": 0.2358,
+      "step": 3860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0674703226791007e-05,
+      "loss": 0.0702,
+      "step": 3861
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0670098574038994e-05,
+      "loss": 0.2644,
+      "step": 3862
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.066549329780546e-05,
+      "loss": 0.225,
+      "step": 3863
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.066088739859679e-05,
+      "loss": 0.1989,
+      "step": 3864
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.065628087691945e-05,
+      "loss": 0.1524,
+      "step": 3865
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0651673733279974e-05,
+      "loss": 0.2455,
+      "step": 3866
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0647065968184956e-05,
+      "loss": 0.1998,
+      "step": 3867
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0642457582141072e-05,
+      "loss": 0.2066,
+      "step": 3868
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.063784857565505e-05,
+      "loss": 0.19,
+      "step": 3869
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0633238949233695e-05,
+      "loss": 0.1899,
+      "step": 3870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.062862870338389e-05,
+      "loss": 0.151,
+      "step": 3871
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.062401783861256e-05,
+      "loss": 0.1639,
+      "step": 3872
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.061940635542672e-05,
+      "loss": 0.2407,
+      "step": 3873
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0614794254333455e-05,
+      "loss": 0.1941,
+      "step": 3874
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0610181535839896e-05,
+      "loss": 0.1852,
+      "step": 3875
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0605568200453265e-05,
+      "loss": 0.196,
+      "step": 3876
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.060095424868084e-05,
+      "loss": 0.178,
+      "step": 3877
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0596339681029968e-05,
+      "loss": 0.2029,
+      "step": 3878
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0591724498008065e-05,
+      "loss": 0.2136,
+      "step": 3879
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0587108700122614e-05,
+      "loss": 0.225,
+      "step": 3880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0582492287881162e-05,
+      "loss": 0.1178,
+      "step": 3881
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.057787526179134e-05,
+      "loss": 0.2017,
+      "step": 3882
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.057325762236082e-05,
+      "loss": 0.1928,
+      "step": 3883
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0568639370097367e-05,
+      "loss": 0.1562,
+      "step": 3884
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0564020505508795e-05,
+      "loss": 0.1329,
+      "step": 3885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0559401029102993e-05,
+      "loss": 0.1399,
+      "step": 3886
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.055478094138792e-05,
+      "loss": 0.2524,
+      "step": 3887
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0550160242871592e-05,
+      "loss": 0.155,
+      "step": 3888
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0545538934062112e-05,
+      "loss": 0.248,
+      "step": 3889
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0540917015467623e-05,
+      "loss": 0.1695,
+      "step": 3890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0536294487596358e-05,
+      "loss": 0.1208,
+      "step": 3891
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.05316713509566e-05,
+      "loss": 0.138,
+      "step": 3892
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0527047606056723e-05,
+      "loss": 0.198,
+      "step": 3893
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.052242325340513e-05,
+      "loss": 0.2107,
+      "step": 3894
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0517798293510326e-05,
+      "loss": 0.2155,
+      "step": 3895
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.051317272688087e-05,
+      "loss": 0.1336,
+      "step": 3896
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.050854655402538e-05,
+      "loss": 0.2148,
+      "step": 3897
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0503919775452555e-05,
+      "loss": 0.1899,
+      "step": 3898
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0499292391671155e-05,
+      "loss": 0.2117,
+      "step": 3899
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0494664403189994e-05,
+      "loss": 0.2362,
+      "step": 3900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0490035810517975e-05,
+      "loss": 0.2785,
+      "step": 3901
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.048540661416405e-05,
+      "loss": 0.1827,
+      "step": 3902
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0480776814637247e-05,
+      "loss": 0.1829,
+      "step": 3903
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0476146412446652e-05,
+      "loss": 0.2614,
+      "step": 3904
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0471515408101426e-05,
+      "loss": 0.2633,
+      "step": 3905
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0466883802110788e-05,
+      "loss": 0.2003,
+      "step": 3906
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0462251594984038e-05,
+      "loss": 0.2484,
+      "step": 3907
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0457618787230515e-05,
+      "loss": 0.2082,
+      "step": 3908
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0452985379359655e-05,
+      "loss": 0.144,
+      "step": 3909
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0448351371880933e-05,
+      "loss": 0.284,
+      "step": 3910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0443716765303916e-05,
+      "loss": 0.1734,
+      "step": 3911
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.043908156013821e-05,
+      "loss": 0.2657,
+      "step": 3912
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0434445756893512e-05,
+      "loss": 0.2088,
+      "step": 3913
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0429809356079566e-05,
+      "loss": 0.1838,
+      "step": 3914
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0425172358206187e-05,
+      "loss": 0.2121,
+      "step": 3915
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.042053476378326e-05,
+      "loss": 0.1913,
+      "step": 3916
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0415896573320736e-05,
+      "loss": 0.1776,
+      "step": 3917
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.041125778732863e-05,
+      "loss": 0.2145,
+      "step": 3918
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0406618406317003e-05,
+      "loss": 0.1838,
+      "step": 3919
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.040197843079602e-05,
+      "loss": 0.2417,
+      "step": 3920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.039733786127588e-05,
+      "loss": 0.2246,
+      "step": 3921
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0392696698266868e-05,
+      "loss": 0.2117,
+      "step": 3922
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.038805494227931e-05,
+      "loss": 0.3229,
+      "step": 3923
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0383412593823623e-05,
+      "loss": 0.2479,
+      "step": 3924
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0378769653410273e-05,
+      "loss": 0.1918,
+      "step": 3925
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0374126121549794e-05,
+      "loss": 0.2097,
+      "step": 3926
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0369481998752787e-05,
+      "loss": 0.2053,
+      "step": 3927
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.036483728552992e-05,
+      "loss": 0.2318,
+      "step": 3928
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0360191982391927e-05,
+      "loss": 0.1807,
+      "step": 3929
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0355546089849597e-05,
+      "loss": 0.1957,
+      "step": 3930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.035089960841379e-05,
+      "loss": 0.209,
+      "step": 3931
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.034625253859544e-05,
+      "loss": 0.1068,
+      "step": 3932
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0341604880905523e-05,
+      "loss": 0.2966,
+      "step": 3933
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0336956635855102e-05,
+      "loss": 0.1496,
+      "step": 3934
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.03323078039553e-05,
+      "loss": 0.2166,
+      "step": 3935
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0327658385717293e-05,
+      "loss": 0.2224,
+      "step": 3936
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0323008381652328e-05,
+      "loss": 0.1802,
+      "step": 3937
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.031835779227173e-05,
+      "loss": 0.1313,
+      "step": 3938
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0313706618086855e-05,
+      "loss": 0.2964,
+      "step": 3939
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.030905485960916e-05,
+      "loss": 0.2372,
+      "step": 3940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0304402517350147e-05,
+      "loss": 0.1674,
+      "step": 3941
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0299749591821384e-05,
+      "loss": 0.1974,
+      "step": 3942
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0295096083534505e-05,
+      "loss": 0.2768,
+      "step": 3943
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0290441993001207e-05,
+      "loss": 0.1295,
+      "step": 3944
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0285787320733246e-05,
+      "loss": 0.2658,
+      "step": 3945
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.028113206724246e-05,
+      "loss": 0.2516,
+      "step": 3946
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0276476233040735e-05,
+      "loss": 0.2126,
+      "step": 3947
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.027181981864002e-05,
+      "loss": 0.1405,
+      "step": 3948
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0267162824552334e-05,
+      "loss": 0.2606,
+      "step": 3949
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0262505251289757e-05,
+      "loss": 0.19,
+      "step": 3950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0257847099364435e-05,
+      "loss": 0.1599,
+      "step": 3951
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.025318836928858e-05,
+      "loss": 0.1555,
+      "step": 3952
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0248529061574457e-05,
+      "loss": 0.1912,
+      "step": 3953
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0243869176734407e-05,
+      "loss": 0.145,
+      "step": 3954
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0239208715280824e-05,
+      "loss": 0.1828,
+      "step": 3955
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0234547677726175e-05,
+      "loss": 0.2603,
+      "step": 3956
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0229886064582982e-05,
+      "loss": 0.1956,
+      "step": 3957
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0225223876363832e-05,
+      "loss": 0.1419,
+      "step": 3958
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0220561113581386e-05,
+      "loss": 0.0901,
+      "step": 3959
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.021589777674835e-05,
+      "loss": 0.1503,
+      "step": 3960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.021123386637751e-05,
+      "loss": 0.1313,
+      "step": 3961
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0206569382981704e-05,
+      "loss": 0.0853,
+      "step": 3962
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0201904327073836e-05,
+      "loss": 0.1724,
+      "step": 3963
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0197238699166867e-05,
+      "loss": 0.197,
+      "step": 3964
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0192572499773844e-05,
+      "loss": 0.2174,
+      "step": 3965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0187905729407844e-05,
+      "loss": 0.1867,
+      "step": 3966
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.018323838858203e-05,
+      "loss": 0.2079,
+      "step": 3967
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.017857047780962e-05,
+      "loss": 0.1712,
+      "step": 3968
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.01739019976039e-05,
+      "loss": 0.2052,
+      "step": 3969
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0169232948478205e-05,
+      "loss": 0.1879,
+      "step": 3970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.016456333094595e-05,
+      "loss": 0.1929,
+      "step": 3971
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0159893145520593e-05,
+      "loss": 0.1763,
+      "step": 3972
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.015522239271568e-05,
+      "loss": 0.2138,
+      "step": 3973
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0150551073044797e-05,
+      "loss": 0.2156,
+      "step": 3974
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0145879187021594e-05,
+      "loss": 0.2211,
+      "step": 3975
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0141206735159803e-05,
+      "loss": 0.2134,
+      "step": 3976
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0136533717973196e-05,
+      "loss": 0.2429,
+      "step": 3977
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.013186013597562e-05,
+      "loss": 0.1545,
+      "step": 3978
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0127185989680975e-05,
+      "loss": 0.1913,
+      "step": 3979
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0122511279603234e-05,
+      "loss": 0.2425,
+      "step": 3980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0117836006256417e-05,
+      "loss": 0.2009,
+      "step": 3981
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.011316017015463e-05,
+      "loss": 0.1978,
+      "step": 3982
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.010848377181201e-05,
+      "loss": 0.1958,
+      "step": 3983
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0103806811742782e-05,
+      "loss": 0.1671,
+      "step": 3984
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.009912929046122e-05,
+      "loss": 0.1971,
+      "step": 3985
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.009445120848166e-05,
+      "loss": 0.2187,
+      "step": 3986
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0089772566318502e-05,
+      "loss": 0.1979,
+      "step": 3987
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0085093364486208e-05,
+      "loss": 0.1355,
+      "step": 3988
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0080413603499303e-05,
+      "loss": 0.2025,
+      "step": 3989
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.007573328387237e-05,
+      "loss": 0.2216,
+      "step": 3990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.007105240612005e-05,
+      "loss": 0.219,
+      "step": 3991
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.006637097075706e-05,
+      "loss": 0.1401,
+      "step": 3992
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0061688978298167e-05,
+      "loss": 0.1302,
+      "step": 3993
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0057006429258196e-05,
+      "loss": 0.2173,
+      "step": 3994
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0052323324152032e-05,
+      "loss": 0.2336,
+      "step": 3995
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0047639663494644e-05,
+      "loss": 0.18,
+      "step": 3996
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0042955447801028e-05,
+      "loss": 0.1143,
+      "step": 3997
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.003827067758627e-05,
+      "loss": 0.2005,
+      "step": 3998
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0033585353365503e-05,
+      "loss": 0.1552,
+      "step": 3999
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0028899475653918e-05,
+      "loss": 0.1422,
+      "step": 4000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.002421304496678e-05,
+      "loss": 0.3215,
+      "step": 4001
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0019526061819397e-05,
+      "loss": 0.1712,
+      "step": 4002
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0014838526727154e-05,
+      "loss": 0.2578,
+      "step": 4003
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.001015044020549e-05,
+      "loss": 0.251,
+      "step": 4004
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0005461802769906e-05,
+      "loss": 0.2131,
+      "step": 4005
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.000077261493596e-05,
+      "loss": 0.2123,
+      "step": 4006
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9996082877219274e-05,
+      "loss": 0.302,
+      "step": 4007
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9991392590135523e-05,
+      "loss": 0.1983,
+      "step": 4008
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9986701754200463e-05,
+      "loss": 0.3313,
+      "step": 4009
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9982010369929886e-05,
+      "loss": 0.1993,
+      "step": 4010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.997731843783966e-05,
+      "loss": 0.1877,
+      "step": 4011
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.99726259584457e-05,
+      "loss": 0.1342,
+      "step": 4012
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9967932932263994e-05,
+      "loss": 0.2496,
+      "step": 4013
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9963239359810586e-05,
+      "loss": 0.2126,
+      "step": 4014
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9958545241601574e-05,
+      "loss": 0.2454,
+      "step": 4015
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.995385057815313e-05,
+      "loss": 0.2029,
+      "step": 4016
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9949155369981473e-05,
+      "loss": 0.1783,
+      "step": 4017
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9944459617602882e-05,
+      "loss": 0.1605,
+      "step": 4018
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9939763321533704e-05,
+      "loss": 0.1354,
+      "step": 4019
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.993506648229034e-05,
+      "loss": 0.1783,
+      "step": 4020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9930369100389257e-05,
+      "loss": 0.1534,
+      "step": 4021
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9925671176346966e-05,
+      "loss": 0.149,
+      "step": 4022
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9920972710680058e-05,
+      "loss": 0.2282,
+      "step": 4023
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9916273703905173e-05,
+      "loss": 0.204,
+      "step": 4024
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9911574156539014e-05,
+      "loss": 0.2173,
+      "step": 4025
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9906874069098332e-05,
+      "loss": 0.1644,
+      "step": 4026
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9902173442099953e-05,
+      "loss": 0.2264,
+      "step": 4027
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9897472276060754e-05,
+      "loss": 0.1654,
+      "step": 4028
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9892770571497678e-05,
+      "loss": 0.1168,
+      "step": 4029
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9888068328927718e-05,
+      "loss": 0.1479,
+      "step": 4030
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.988336554886793e-05,
+      "loss": 0.1013,
+      "step": 4031
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.987866223183543e-05,
+      "loss": 0.245,
+      "step": 4032
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9873958378347393e-05,
+      "loss": 0.1624,
+      "step": 4033
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9869253988921053e-05,
+      "loss": 0.1679,
+      "step": 4034
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9864549064073706e-05,
+      "loss": 0.1203,
+      "step": 4035
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9859843604322697e-05,
+      "loss": 0.2426,
+      "step": 4036
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.985513761018544e-05,
+      "loss": 0.2786,
+      "step": 4037
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.98504310821794e-05,
+      "loss": 0.2271,
+      "step": 4038
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.984572402082211e-05,
+      "loss": 0.1007,
+      "step": 4039
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9841016426631155e-05,
+      "loss": 0.2048,
+      "step": 4040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.983630830012418e-05,
+      "loss": 0.1622,
+      "step": 4041
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.983159964181889e-05,
+      "loss": 0.2026,
+      "step": 4042
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9826890452233043e-05,
+      "loss": 0.2579,
+      "step": 4043
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.982218073188446e-05,
+      "loss": 0.2526,
+      "step": 4044
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9817470481291018e-05,
+      "loss": 0.1689,
+      "step": 4045
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.981275970097067e-05,
+      "loss": 0.1282,
+      "step": 4046
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9808048391441384e-05,
+      "loss": 0.2029,
+      "step": 4047
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9803336553221232e-05,
+      "loss": 0.2449,
+      "step": 4048
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9798624186828325e-05,
+      "loss": 0.2209,
+      "step": 4049
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9793911292780828e-05,
+      "loss": 0.1474,
+      "step": 4050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9789197871596966e-05,
+      "loss": 0.1641,
+      "step": 4051
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9784483923795037e-05,
+      "loss": 0.189,
+      "step": 4052
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.977976944989337e-05,
+      "loss": 0.1078,
+      "step": 4053
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.977505445041037e-05,
+      "loss": 0.1311,
+      "step": 4054
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9770338925864505e-05,
+      "loss": 0.1691,
+      "step": 4055
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9765622876774283e-05,
+      "loss": 0.2001,
+      "step": 4056
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9760906303658285e-05,
+      "loss": 0.2271,
+      "step": 4057
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.975618920703514e-05,
+      "loss": 0.2043,
+      "step": 4058
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9751471587423532e-05,
+      "loss": 0.1426,
+      "step": 4059
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.974675344534222e-05,
+      "loss": 0.2625,
+      "step": 4060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.974203478131e-05,
+      "loss": 0.2051,
+      "step": 4061
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.973731559584574e-05,
+      "loss": 0.1372,
+      "step": 4062
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9732595889468354e-05,
+      "loss": 0.1465,
+      "step": 4063
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9727875662696822e-05,
+      "loss": 0.1738,
+      "step": 4064
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9723154916050176e-05,
+      "loss": 0.1945,
+      "step": 4065
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.971843365004751e-05,
+      "loss": 0.1225,
+      "step": 4066
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9713711865207975e-05,
+      "loss": 0.2302,
+      "step": 4067
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.970898956205077e-05,
+      "loss": 0.1509,
+      "step": 4068
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.970426674109516e-05,
+      "loss": 0.2264,
+      "step": 4069
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9699543402860458e-05,
+      "loss": 0.1392,
+      "step": 4070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.969481954786606e-05,
+      "loss": 0.1569,
+      "step": 4071
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9690095176631373e-05,
+      "loss": 0.1694,
+      "step": 4072
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9685370289675904e-05,
+      "loss": 0.1275,
+      "step": 4073
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9680644887519194e-05,
+      "loss": 0.1888,
+      "step": 4074
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9675918970680847e-05,
+      "loss": 0.1326,
+      "step": 4075
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9671192539680524e-05,
+      "loss": 0.2166,
+      "step": 4076
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9666465595037938e-05,
+      "loss": 0.1964,
+      "step": 4077
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9661738137272862e-05,
+      "loss": 0.2504,
+      "step": 4078
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9657010166905127e-05,
+      "loss": 0.1509,
+      "step": 4079
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9652281684454622e-05,
+      "loss": 0.2432,
+      "step": 4080
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9647552690441282e-05,
+      "loss": 0.2614,
+      "step": 4081
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9642823185385108e-05,
+      "loss": 0.1363,
+      "step": 4082
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9638093169806157e-05,
+      "loss": 0.2099,
+      "step": 4083
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9633362644224532e-05,
+      "loss": 0.1167,
+      "step": 4084
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.962863160916041e-05,
+      "loss": 0.202,
+      "step": 4085
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9623900065134e-05,
+      "loss": 0.1934,
+      "step": 4086
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9619168012665595e-05,
+      "loss": 0.217,
+      "step": 4087
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.961443545227552e-05,
+      "loss": 0.2682,
+      "step": 4088
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.960970238448417e-05,
+      "loss": 0.1274,
+      "step": 4089
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9604968809811987e-05,
+      "loss": 0.1019,
+      "step": 4090
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9600234728779476e-05,
+      "loss": 0.1851,
+      "step": 4091
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.959550014190719e-05,
+      "loss": 0.2374,
+      "step": 4092
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9590765049715745e-05,
+      "loss": 0.1861,
+      "step": 4093
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9586029452725813e-05,
+      "loss": 0.2053,
+      "step": 4094
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9581293351458116e-05,
+      "loss": 0.117,
+      "step": 4095
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.957655674643343e-05,
+      "loss": 0.2568,
+      "step": 4096
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9571819638172593e-05,
+      "loss": 0.203,
+      "step": 4097
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9567082027196495e-05,
+      "loss": 0.1378,
+      "step": 4098
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9562343914026086e-05,
+      "loss": 0.1405,
+      "step": 4099
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9557605299182355e-05,
+      "loss": 0.1548,
+      "step": 4100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9552866183186373e-05,
+      "loss": 0.1949,
+      "step": 4101
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.954812656655924e-05,
+      "loss": 0.1754,
+      "step": 4102
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.954338644982212e-05,
+      "loss": 0.2691,
+      "step": 4103
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.953864583349625e-05,
+      "loss": 0.1152,
+      "step": 4104
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9533904718102885e-05,
+      "loss": 0.2175,
+      "step": 4105
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9529163104163373e-05,
+      "loss": 0.1777,
+      "step": 4106
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9524420992199094e-05,
+      "loss": 0.142,
+      "step": 4107
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9519678382731483e-05,
+      "loss": 0.1531,
+      "step": 4108
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9514935276282043e-05,
+      "loss": 0.2465,
+      "step": 4109
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.951019167337233e-05,
+      "loss": 0.1362,
+      "step": 4110
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.950544757452393e-05,
+      "loss": 0.2063,
+      "step": 4111
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9500702980258507e-05,
+      "loss": 0.1602,
+      "step": 4112
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9495957891097787e-05,
+      "loss": 0.1855,
+      "step": 4113
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9491212307563528e-05,
+      "loss": 0.292,
+      "step": 4114
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.948646623017756e-05,
+      "loss": 0.1754,
+      "step": 4115
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9481719659461742e-05,
+      "loss": 0.1216,
+      "step": 4116
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9476972595938023e-05,
+      "loss": 0.1895,
+      "step": 4117
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9472225040128384e-05,
+      "loss": 0.1901,
+      "step": 4118
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9467476992554858e-05,
+      "loss": 0.1519,
+      "step": 4119
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9462728453739543e-05,
+      "loss": 0.2509,
+      "step": 4120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9457979424204588e-05,
+      "loss": 0.2124,
+      "step": 4121
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.945322990447219e-05,
+      "loss": 0.2025,
+      "step": 4122
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.94484798950646e-05,
+      "loss": 0.1877,
+      "step": 4123
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.944372939650414e-05,
+      "loss": 0.2536,
+      "step": 4124
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.943897840931316e-05,
+      "loss": 0.2319,
+      "step": 4125
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9434226934014085e-05,
+      "loss": 0.2256,
+      "step": 4126
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.942947497112938e-05,
+      "loss": 0.1817,
+      "step": 4127
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9424722521181572e-05,
+      "loss": 0.1681,
+      "step": 4128
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9419969584693236e-05,
+      "loss": 0.2684,
+      "step": 4129
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9415216162187e-05,
+      "loss": 0.1025,
+      "step": 4130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9410462254185557e-05,
+      "loss": 0.1658,
+      "step": 4131
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9405707861211634e-05,
+      "loss": 0.1513,
+      "step": 4132
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.940095298378803e-05,
+      "loss": 0.1183,
+      "step": 4133
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9396197622437588e-05,
+      "loss": 0.1444,
+      "step": 4134
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9391441777683206e-05,
+      "loss": 0.2161,
+      "step": 4135
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9386685450047827e-05,
+      "loss": 0.1296,
+      "step": 4136
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.938192864005446e-05,
+      "loss": 0.1559,
+      "step": 4137
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9377171348226167e-05,
+      "loss": 0.112,
+      "step": 4138
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.937241357508605e-05,
+      "loss": 0.1609,
+      "step": 4139
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9367655321157276e-05,
+      "loss": 0.2043,
+      "step": 4140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9362896586963058e-05,
+      "loss": 0.2233,
+      "step": 4141
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.935813737302666e-05,
+      "loss": 0.1908,
+      "step": 4142
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.935337767987142e-05,
+      "loss": 0.2127,
+      "step": 4143
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9348617508020685e-05,
+      "loss": 0.158,
+      "step": 4144
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9343856857997903e-05,
+      "loss": 0.1827,
+      "step": 4145
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9339095730326548e-05,
+      "loss": 0.2039,
+      "step": 4146
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.933433412553015e-05,
+      "loss": 0.1989,
+      "step": 4147
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9329572044132286e-05,
+      "loss": 0.2343,
+      "step": 4148
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.932480948665661e-05,
+      "loss": 0.311,
+      "step": 4149
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.932004645362679e-05,
+      "loss": 0.1545,
+      "step": 4150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.931528294556658e-05,
+      "loss": 0.1964,
+      "step": 4151
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9310518962999774e-05,
+      "loss": 0.2595,
+      "step": 4152
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9305754506450206e-05,
+      "loss": 0.2151,
+      "step": 4153
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9300989576441788e-05,
+      "loss": 0.1837,
+      "step": 4154
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.929622417349846e-05,
+      "loss": 0.1356,
+      "step": 4155
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.929145829814422e-05,
+      "loss": 0.1918,
+      "step": 4156
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.928669195090313e-05,
+      "loss": 0.1849,
+      "step": 4157
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9281925132299297e-05,
+      "loss": 0.1398,
+      "step": 4158
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.927715784285687e-05,
+      "loss": 0.2366,
+      "step": 4159
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.927239008310007e-05,
+      "loss": 0.2109,
+      "step": 4160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.926762185355314e-05,
+      "loss": 0.2028,
+      "step": 4161
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9262853154740413e-05,
+      "loss": 0.1559,
+      "step": 4162
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9258083987186235e-05,
+      "loss": 0.1812,
+      "step": 4163
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9253314351415033e-05,
+      "loss": 0.1923,
+      "step": 4164
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9248544247951268e-05,
+      "loss": 0.1736,
+      "step": 4165
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9243773677319464e-05,
+      "loss": 0.1285,
+      "step": 4166
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.923900264004418e-05,
+      "loss": 0.1787,
+      "step": 4167
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9234231136650054e-05,
+      "loss": 0.1943,
+      "step": 4168
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.922945916766174e-05,
+      "loss": 0.1238,
+      "step": 4169
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9224686733603974e-05,
+      "loss": 0.1585,
+      "step": 4170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.921991383500153e-05,
+      "loss": 0.2256,
+      "step": 4171
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9215140472379227e-05,
+      "loss": 0.2041,
+      "step": 4172
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9210366646261943e-05,
+      "loss": 0.2589,
+      "step": 4173
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.920559235717462e-05,
+      "loss": 0.1912,
+      "step": 4174
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9200817605642212e-05,
+      "loss": 0.1985,
+      "step": 4175
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9196042392189772e-05,
+      "loss": 0.1526,
+      "step": 4176
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9191266717342368e-05,
+      "loss": 0.1419,
+      "step": 4177
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9186490581625134e-05,
+      "loss": 0.1631,
+      "step": 4178
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.918171398556325e-05,
+      "loss": 0.3077,
+      "step": 4179
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9176936929681948e-05,
+      "loss": 0.1578,
+      "step": 4180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9172159414506513e-05,
+      "loss": 0.1783,
+      "step": 4181
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.916738144056228e-05,
+      "loss": 0.2009,
+      "step": 4182
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.916260300837463e-05,
+      "loss": 0.074,
+      "step": 4183
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9157824118468996e-05,
+      "loss": 0.2098,
+      "step": 4184
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9153044771370867e-05,
+      "loss": 0.2133,
+      "step": 4185
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9148264967605776e-05,
+      "loss": 0.1683,
+      "step": 4186
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9143484707699304e-05,
+      "loss": 0.1949,
+      "step": 4187
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9138703992177093e-05,
+      "loss": 0.2646,
+      "step": 4188
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9133922821564828e-05,
+      "loss": 0.1916,
+      "step": 4189
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.912914119638824e-05,
+      "loss": 0.2526,
+      "step": 4190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9124359117173114e-05,
+      "loss": 0.2118,
+      "step": 4191
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.911957658444529e-05,
+      "loss": 0.1998,
+      "step": 4192
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9114793598730653e-05,
+      "loss": 0.2256,
+      "step": 4193
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9110010160555128e-05,
+      "loss": 0.257,
+      "step": 4194
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9105226270444714e-05,
+      "loss": 0.2653,
+      "step": 4195
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9100441928925437e-05,
+      "loss": 0.1712,
+      "step": 4196
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9095657136523378e-05,
+      "loss": 0.2016,
+      "step": 4197
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.909087189376468e-05,
+      "loss": 0.2209,
+      "step": 4198
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9086086201175524e-05,
+      "loss": 0.1789,
+      "step": 4199
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9081300059282135e-05,
+      "loss": 0.1412,
+      "step": 4200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9076513468610805e-05,
+      "loss": 0.1477,
+      "step": 4201
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.907172642968786e-05,
+      "loss": 0.2362,
+      "step": 4202
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9066938943039682e-05,
+      "loss": 0.2039,
+      "step": 4203
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.90621510091927e-05,
+      "loss": 0.1848,
+      "step": 4204
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9057362628673394e-05,
+      "loss": 0.1578,
+      "step": 4205
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9052573802008285e-05,
+      "loss": 0.2123,
+      "step": 4206
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9047784529723967e-05,
+      "loss": 0.1996,
+      "step": 4207
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9042994812347047e-05,
+      "loss": 0.0924,
+      "step": 4208
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9038204650404214e-05,
+      "loss": 0.2162,
+      "step": 4209
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.903341404442219e-05,
+      "loss": 0.2192,
+      "step": 4210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9028622994927742e-05,
+      "loss": 0.2,
+      "step": 4211
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9023831502447693e-05,
+      "loss": 0.1544,
+      "step": 4212
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9019039567508922e-05,
+      "loss": 0.1649,
+      "step": 4213
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9014247190638338e-05,
+      "loss": 0.1315,
+      "step": 4214
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.900945437236291e-05,
+      "loss": 0.2012,
+      "step": 4215
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9004661113209658e-05,
+      "loss": 0.2016,
+      "step": 4216
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8999867413705644e-05,
+      "loss": 0.1371,
+      "step": 4217
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8995073274377987e-05,
+      "loss": 0.1749,
+      "step": 4218
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8990278695753835e-05,
+      "loss": 0.191,
+      "step": 4219
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.898548367836041e-05,
+      "loss": 0.21,
+      "step": 4220
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.898068822272497e-05,
+      "loss": 0.1263,
+      "step": 4221
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8975892329374813e-05,
+      "loss": 0.1365,
+      "step": 4222
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8971095998837296e-05,
+      "loss": 0.1824,
+      "step": 4223
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896629923163983e-05,
+      "loss": 0.1845,
+      "step": 4224
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8961502028309854e-05,
+      "loss": 0.1555,
+      "step": 4225
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8956704389374868e-05,
+      "loss": 0.2171,
+      "step": 4226
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.895190631536243e-05,
+      "loss": 0.2363,
+      "step": 4227
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8947107806800115e-05,
+      "loss": 0.1604,
+      "step": 4228
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.894230886421558e-05,
+      "loss": 0.1982,
+      "step": 4229
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.893750948813651e-05,
+      "loss": 0.263,
+      "step": 4230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.893270967909064e-05,
+      "loss": 0.2618,
+      "step": 4231
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892790943760576e-05,
+      "loss": 0.2131,
+      "step": 4232
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8923108764209693e-05,
+      "loss": 0.1719,
+      "step": 4233
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8918307659430332e-05,
+      "loss": 0.1472,
+      "step": 4234
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8913506123795594e-05,
+      "loss": 0.182,
+      "step": 4235
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8908704157833458e-05,
+      "loss": 0.2638,
+      "step": 4236
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.890390176207194e-05,
+      "loss": 0.2177,
+      "step": 4237
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8899098937039127e-05,
+      "loss": 0.1858,
+      "step": 4238
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.889429568326311e-05,
+      "loss": 0.2095,
+      "step": 4239
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.888949200127207e-05,
+      "loss": 0.1226,
+      "step": 4240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.888468789159422e-05,
+      "loss": 0.1746,
+      "step": 4241
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8879883354757804e-05,
+      "loss": 0.2454,
+      "step": 4242
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8875078391291137e-05,
+      "loss": 0.2585,
+      "step": 4243
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8870273001722564e-05,
+      "loss": 0.1185,
+      "step": 4244
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8865467186580488e-05,
+      "loss": 0.1946,
+      "step": 4245
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8860660946393362e-05,
+      "loss": 0.1888,
+      "step": 4246
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8855854281689657e-05,
+      "loss": 0.2057,
+      "step": 4247
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.885104719299793e-05,
+      "loss": 0.1614,
+      "step": 4248
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8846239680846763e-05,
+      "loss": 0.1038,
+      "step": 4249
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.884143174576478e-05,
+      "loss": 0.2192,
+      "step": 4250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8836623388280667e-05,
+      "loss": 0.1831,
+      "step": 4251
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.883181460892315e-05,
+      "loss": 0.1567,
+      "step": 4252
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.882700540822099e-05,
+      "loss": 0.1695,
+      "step": 4253
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.882219578670302e-05,
+      "loss": 0.1799,
+      "step": 4254
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.881738574489809e-05,
+      "loss": 0.1216,
+      "step": 4255
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.881257528333511e-05,
+      "loss": 0.2227,
+      "step": 4256
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8807764402543058e-05,
+      "loss": 0.221,
+      "step": 4257
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8802953103050903e-05,
+      "loss": 0.1861,
+      "step": 4258
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.879814138538772e-05,
+      "loss": 0.1647,
+      "step": 4259
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8793329250082593e-05,
+      "loss": 0.1573,
+      "step": 4260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8788516697664662e-05,
+      "loss": 0.1497,
+      "step": 4261
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.878370372866311e-05,
+      "loss": 0.176,
+      "step": 4262
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.877889034360718e-05,
+      "loss": 0.1584,
+      "step": 4263
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8774076543026136e-05,
+      "loss": 0.1308,
+      "step": 4264
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.876926232744931e-05,
+      "loss": 0.2213,
+      "step": 4265
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8764447697406073e-05,
+      "loss": 0.1932,
+      "step": 4266
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8759632653425834e-05,
+      "loss": 0.1824,
+      "step": 4267
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8754817196038053e-05,
+      "loss": 0.1457,
+      "step": 4268
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.875000132577224e-05,
+      "loss": 0.257,
+      "step": 4269
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.874518504315794e-05,
+      "loss": 0.2405,
+      "step": 4270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8740368348724752e-05,
+      "loss": 0.2254,
+      "step": 4271
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.873555124300232e-05,
+      "loss": 0.2037,
+      "step": 4272
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.873073372652033e-05,
+      "loss": 0.176,
+      "step": 4273
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.872591579980851e-05,
+      "loss": 0.2191,
+      "step": 4274
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8721097463396643e-05,
+      "loss": 0.2114,
+      "step": 4275
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8716278717814546e-05,
+      "loss": 0.1564,
+      "step": 4276
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.871145956359209e-05,
+      "loss": 0.1507,
+      "step": 4277
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8706640001259186e-05,
+      "loss": 0.2135,
+      "step": 4278
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8701820031345788e-05,
+      "loss": 0.1457,
+      "step": 4279
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8696999654381905e-05,
+      "loss": 0.1464,
+      "step": 4280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8692178870897572e-05,
+      "loss": 0.157,
+      "step": 4281
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8687357681422894e-05,
+      "loss": 0.1551,
+      "step": 4282
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.868253608648799e-05,
+      "loss": 0.2161,
+      "step": 4283
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.867771408662306e-05,
+      "loss": 0.1026,
+      "step": 4284
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.867289168235831e-05,
+      "loss": 0.1784,
+      "step": 4285
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8668068874224027e-05,
+      "loss": 0.1915,
+      "step": 4286
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8663245662750512e-05,
+      "loss": 0.1118,
+      "step": 4287
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.865842204846813e-05,
+      "loss": 0.2044,
+      "step": 4288
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8653598031907275e-05,
+      "loss": 0.1945,
+      "step": 4289
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8648773613598405e-05,
+      "loss": 0.0771,
+      "step": 4290
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8643948794072006e-05,
+      "loss": 0.2819,
+      "step": 4291
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8639123573858613e-05,
+      "loss": 0.1983,
+      "step": 4292
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8634297953488804e-05,
+      "loss": 0.0896,
+      "step": 4293
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.86294719334932e-05,
+      "loss": 0.2042,
+      "step": 4294
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8624645514402473e-05,
+      "loss": 0.2286,
+      "step": 4295
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8619818696747337e-05,
+      "loss": 0.1143,
+      "step": 4296
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8614991481058535e-05,
+      "loss": 0.2272,
+      "step": 4297
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.861016386786687e-05,
+      "loss": 0.2321,
+      "step": 4298
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8605335857703193e-05,
+      "loss": 0.2726,
+      "step": 4299
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.860050745109838e-05,
+      "loss": 0.2148,
+      "step": 4300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.859567864858336e-05,
+      "loss": 0.1522,
+      "step": 4301
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8590849450689113e-05,
+      "loss": 0.1868,
+      "step": 4302
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8586019857946648e-05,
+      "loss": 0.296,
+      "step": 4303
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8581189870887036e-05,
+      "loss": 0.1079,
+      "step": 4304
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.857635949004137e-05,
+      "loss": 0.2126,
+      "step": 4305
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.85715287159408e-05,
+      "loss": 0.1819,
+      "step": 4306
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8566697549116513e-05,
+      "loss": 0.1819,
+      "step": 4307
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.856186599009974e-05,
+      "loss": 0.2033,
+      "step": 4308
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.855703403942177e-05,
+      "loss": 0.1478,
+      "step": 4309
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.855220169761391e-05,
+      "loss": 0.1619,
+      "step": 4310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.854736896520753e-05,
+      "loss": 0.2726,
+      "step": 4311
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.854253584273402e-05,
+      "loss": 0.2949,
+      "step": 4312
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.853770233072485e-05,
+      "loss": 0.1972,
+      "step": 4313
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8532868429711492e-05,
+      "loss": 0.1998,
+      "step": 4314
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8528034140225494e-05,
+      "loss": 0.2352,
+      "step": 4315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8523199462798422e-05,
+      "loss": 0.2577,
+      "step": 4316
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.85183643979619e-05,
+      "loss": 0.1165,
+      "step": 4317
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.851352894624759e-05,
+      "loss": 0.1476,
+      "step": 4318
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8508693108187194e-05,
+      "loss": 0.1449,
+      "step": 4319
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8503856884312453e-05,
+      "loss": 0.191,
+      "step": 4320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8499020275155168e-05,
+      "loss": 0.1331,
+      "step": 4321
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.849418328124716e-05,
+      "loss": 0.2047,
+      "step": 4322
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8489345903120314e-05,
+      "loss": 0.2416,
+      "step": 4323
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8484508141306533e-05,
+      "loss": 0.182,
+      "step": 4324
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8479669996337785e-05,
+      "loss": 0.1567,
+      "step": 4325
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8474831468746057e-05,
+      "loss": 0.1685,
+      "step": 4326
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8469992559063415e-05,
+      "loss": 0.1977,
+      "step": 4327
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.846515326782192e-05,
+      "loss": 0.169,
+      "step": 4328
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.846031359555371e-05,
+      "loss": 0.133,
+      "step": 4329
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.845547354279095e-05,
+      "loss": 0.1228,
+      "step": 4330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8450633110065853e-05,
+      "loss": 0.2255,
+      "step": 4331
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8445792297910666e-05,
+      "loss": 0.155,
+      "step": 4332
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8440951106857687e-05,
+      "loss": 0.247,
+      "step": 4333
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8436109537439246e-05,
+      "loss": 0.2971,
+      "step": 4334
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.843126759018773e-05,
+      "loss": 0.2251,
+      "step": 4335
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.842642526563554e-05,
+      "loss": 0.1724,
+      "step": 4336
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8421582564315156e-05,
+      "loss": 0.2363,
+      "step": 4337
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8416739486759067e-05,
+      "loss": 0.1454,
+      "step": 4338
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8411896033499816e-05,
+      "loss": 0.1668,
+      "step": 4339
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.840705220506999e-05,
+      "loss": 0.1518,
+      "step": 4340
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8402208002002216e-05,
+      "loss": 0.1603,
+      "step": 4341
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8397363424829158e-05,
+      "loss": 0.1496,
+      "step": 4342
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8392518474083524e-05,
+      "loss": 0.2359,
+      "step": 4343
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.838767315029806e-05,
+      "loss": 0.1685,
+      "step": 4344
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8382827454005557e-05,
+      "loss": 0.2025,
+      "step": 4345
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8377981385738852e-05,
+      "loss": 0.2056,
+      "step": 4346
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8373134946030806e-05,
+      "loss": 0.222,
+      "step": 4347
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8368288135414336e-05,
+      "loss": 0.1705,
+      "step": 4348
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.83634409544224e-05,
+      "loss": 0.1926,
+      "step": 4349
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8358593403587982e-05,
+      "loss": 0.226,
+      "step": 4350
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8353745483444122e-05,
+      "loss": 0.1404,
+      "step": 4351
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8348897194523903e-05,
+      "loss": 0.1531,
+      "step": 4352
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8344048537360423e-05,
+      "loss": 0.1578,
+      "step": 4353
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8339199512486856e-05,
+      "loss": 0.1057,
+      "step": 4354
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8334350120436388e-05,
+      "loss": 0.1539,
+      "step": 4355
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8329500361742257e-05,
+      "loss": 0.158,
+      "step": 4356
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8324650236937742e-05,
+      "loss": 0.2175,
+      "step": 4357
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8319799746556162e-05,
+      "loss": 0.195,
+      "step": 4358
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.831494889113087e-05,
+      "loss": 0.1559,
+      "step": 4359
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.831009767119527e-05,
+      "loss": 0.1354,
+      "step": 4360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8305246087282793e-05,
+      "loss": 0.3121,
+      "step": 4361
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8300394139926928e-05,
+      "loss": 0.2555,
+      "step": 4362
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8295541829661182e-05,
+      "loss": 0.1651,
+      "step": 4363
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.829068915701912e-05,
+      "loss": 0.1881,
+      "step": 4364
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8285836122534333e-05,
+      "loss": 0.1865,
+      "step": 4365
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.828098272674047e-05,
+      "loss": 0.2622,
+      "step": 4366
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8276128970171188e-05,
+      "loss": 0.1392,
+      "step": 4367
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8271274853360224e-05,
+      "loss": 0.1746,
+      "step": 4368
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.826642037684133e-05,
+      "loss": 0.233,
+      "step": 4369
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8261565541148292e-05,
+      "loss": 0.1699,
+      "step": 4370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8256710346814962e-05,
+      "loss": 0.1866,
+      "step": 4371
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8251854794375194e-05,
+      "loss": 0.2045,
+      "step": 4372
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8246998884362924e-05,
+      "loss": 0.1871,
+      "step": 4373
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.824214261731209e-05,
+      "loss": 0.2479,
+      "step": 4374
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8237285993756694e-05,
+      "loss": 0.2111,
+      "step": 4375
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.823242901423076e-05,
+      "loss": 0.1671,
+      "step": 4376
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.822757167926837e-05,
+      "loss": 0.1704,
+      "step": 4377
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8222713989403624e-05,
+      "loss": 0.2426,
+      "step": 4378
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.821785594517068e-05,
+      "loss": 0.1903,
+      "step": 4379
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8212997547103722e-05,
+      "loss": 0.1099,
+      "step": 4380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8208138795736978e-05,
+      "loss": 0.1657,
+      "step": 4381
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8203279691604714e-05,
+      "loss": 0.1513,
+      "step": 4382
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8198420235241232e-05,
+      "loss": 0.1535,
+      "step": 4383
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8193560427180882e-05,
+      "loss": 0.2086,
+      "step": 4384
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8188700267958044e-05,
+      "loss": 0.2219,
+      "step": 4385
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8183839758107133e-05,
+      "loss": 0.2086,
+      "step": 4386
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8178978898162617e-05,
+      "loss": 0.142,
+      "step": 4387
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8174117688658988e-05,
+      "loss": 0.1553,
+      "step": 4388
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8169256130130785e-05,
+      "loss": 0.1735,
+      "step": 4389
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8164394223112583e-05,
+      "loss": 0.1272,
+      "step": 4390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8159531968138997e-05,
+      "loss": 0.2118,
+      "step": 4391
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8154669365744672e-05,
+      "loss": 0.1915,
+      "step": 4392
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8149806416464304e-05,
+      "loss": 0.196,
+      "step": 4393
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.814494312083262e-05,
+      "loss": 0.2144,
+      "step": 4394
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.814007947938438e-05,
+      "loss": 0.1707,
+      "step": 4395
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.81352154926544e-05,
+      "loss": 0.1858,
+      "step": 4396
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.813035116117751e-05,
+      "loss": 0.21,
+      "step": 4397
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.812548648548859e-05,
+      "loss": 0.128,
+      "step": 4398
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8120621466122564e-05,
+      "loss": 0.2282,
+      "step": 4399
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8115756103614383e-05,
+      "loss": 0.1143,
+      "step": 4400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8110890398499043e-05,
+      "loss": 0.1561,
+      "step": 4401
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.810602435131158e-05,
+      "loss": 0.1476,
+      "step": 4402
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8101157962587047e-05,
+      "loss": 0.1642,
+      "step": 4403
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.809629123286056e-05,
+      "loss": 0.2133,
+      "step": 4404
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8091424162667267e-05,
+      "loss": 0.1823,
+      "step": 4405
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.808655675254234e-05,
+      "loss": 0.204,
+      "step": 4406
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8081689003020997e-05,
+      "loss": 0.2611,
+      "step": 4407
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8076820914638503e-05,
+      "loss": 0.1517,
+      "step": 4408
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.807195248793014e-05,
+      "loss": 0.1511,
+      "step": 4409
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.806708372343125e-05,
+      "loss": 0.1642,
+      "step": 4410
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8062214621677185e-05,
+      "loss": 0.1466,
+      "step": 4411
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8057345183203364e-05,
+      "loss": 0.1971,
+      "step": 4412
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.805247540854522e-05,
+      "loss": 0.1412,
+      "step": 4413
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.804760529823823e-05,
+      "loss": 0.1503,
+      "step": 4414
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.804273485281791e-05,
+      "loss": 0.2036,
+      "step": 4415
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8037864072819823e-05,
+      "loss": 0.1902,
+      "step": 4416
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8032992958779544e-05,
+      "loss": 0.2408,
+      "step": 4417
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8028121511232704e-05,
+      "loss": 0.1461,
+      "step": 4418
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8023249730714966e-05,
+      "loss": 0.2004,
+      "step": 4419
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8018377617762024e-05,
+      "loss": 0.2249,
+      "step": 4420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.801350517290962e-05,
+      "loss": 0.1156,
+      "step": 4421
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8008632396693522e-05,
+      "loss": 0.1625,
+      "step": 4422
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8003759289649537e-05,
+      "loss": 0.2076,
+      "step": 4423
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7998885852313517e-05,
+      "loss": 0.1034,
+      "step": 4424
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7994012085221334e-05,
+      "loss": 0.1509,
+      "step": 4425
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.798913798890891e-05,
+      "loss": 0.1741,
+      "step": 4426
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7984263563912197e-05,
+      "loss": 0.1698,
+      "step": 4427
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7979388810767184e-05,
+      "loss": 0.0758,
+      "step": 4428
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.79745137300099e-05,
+      "loss": 0.2114,
+      "step": 4429
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7969638322176404e-05,
+      "loss": 0.159,
+      "step": 4430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7964762587802797e-05,
+      "loss": 0.2077,
+      "step": 4431
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.795988652742521e-05,
+      "loss": 0.1975,
+      "step": 4432
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7955010141579806e-05,
+      "loss": 0.1989,
+      "step": 4433
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.79501334308028e-05,
+      "loss": 0.189,
+      "step": 4434
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7945256395630433e-05,
+      "loss": 0.0989,
+      "step": 4435
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.794037903659897e-05,
+      "loss": 0.172,
+      "step": 4436
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.793550135424474e-05,
+      "loss": 0.2451,
+      "step": 4437
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7930623349104077e-05,
+      "loss": 0.172,
+      "step": 4438
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7925745021713374e-05,
+      "loss": 0.0992,
+      "step": 4439
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7920866372609038e-05,
+      "loss": 0.1189,
+      "step": 4440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7915987402327538e-05,
+      "loss": 0.2307,
+      "step": 4441
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.791110811140535e-05,
+      "loss": 0.1588,
+      "step": 4442
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.790622850037901e-05,
+      "loss": 0.1764,
+      "step": 4443
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7901348569785066e-05,
+      "loss": 0.1561,
+      "step": 4444
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7896468320160124e-05,
+      "loss": 0.1562,
+      "step": 4445
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7891587752040805e-05,
+      "loss": 0.1286,
+      "step": 4446
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.788670686596378e-05,
+      "loss": 0.2146,
+      "step": 4447
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7881825662465746e-05,
+      "loss": 0.2411,
+      "step": 4448
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7876944142083446e-05,
+      "loss": 0.1372,
+      "step": 4449
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7872062305353637e-05,
+      "loss": 0.2883,
+      "step": 4450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7867180152813132e-05,
+      "loss": 0.2112,
+      "step": 4451
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.786229768499877e-05,
+      "loss": 0.2026,
+      "step": 4452
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.785741490244742e-05,
+      "loss": 0.1523,
+      "step": 4453
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.785253180569599e-05,
+      "loss": 0.1857,
+      "step": 4454
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.784764839528144e-05,
+      "loss": 0.1904,
+      "step": 4455
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7842764671740722e-05,
+      "loss": 0.222,
+      "step": 4456
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7837880635610863e-05,
+      "loss": 0.1967,
+      "step": 4457
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7832996287428914e-05,
+      "loss": 0.1568,
+      "step": 4458
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.782811162773194e-05,
+      "loss": 0.187,
+      "step": 4459
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.782322665705707e-05,
+      "loss": 0.2303,
+      "step": 4460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.781834137594145e-05,
+      "loss": 0.1897,
+      "step": 4461
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.781345578492226e-05,
+      "loss": 0.1945,
+      "step": 4462
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7808569884536716e-05,
+      "loss": 0.2139,
+      "step": 4463
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7803683675322073e-05,
+      "loss": 0.1511,
+      "step": 4464
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7798797157815613e-05,
+      "loss": 0.2168,
+      "step": 4465
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7793910332554666e-05,
+      "loss": 0.1662,
+      "step": 4466
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.778902320007657e-05,
+      "loss": 0.2662,
+      "step": 4467
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7784135760918723e-05,
+      "loss": 0.2366,
+      "step": 4468
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.777924801561854e-05,
+      "loss": 0.1556,
+      "step": 4469
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.777435996471348e-05,
+      "loss": 0.1365,
+      "step": 4470
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7769471608741024e-05,
+      "loss": 0.186,
+      "step": 4471
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.77645829482387e-05,
+      "loss": 0.1951,
+      "step": 4472
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7759693983744058e-05,
+      "loss": 0.1618,
+      "step": 4473
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7754804715794695e-05,
+      "loss": 0.1161,
+      "step": 4474
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7749915144928213e-05,
+      "loss": 0.1454,
+      "step": 4475
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7745025271682293e-05,
+      "loss": 0.1214,
+      "step": 4476
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7740135096594606e-05,
+      "loss": 0.2457,
+      "step": 4477
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.773524462020288e-05,
+      "loss": 0.2365,
+      "step": 4478
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7730353843044866e-05,
+      "loss": 0.1861,
+      "step": 4479
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7725462765658358e-05,
+      "loss": 0.1584,
+      "step": 4480
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7720571388581164e-05,
+      "loss": 0.1776,
+      "step": 4481
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.771567971235115e-05,
+      "loss": 0.1585,
+      "step": 4482
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7710787737506204e-05,
+      "loss": 0.1777,
+      "step": 4483
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.770589546458423e-05,
+      "loss": 0.1668,
+      "step": 4484
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.77010028941232e-05,
+      "loss": 0.2369,
+      "step": 4485
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7696110026661086e-05,
+      "loss": 0.191,
+      "step": 4486
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.76912168627359e-05,
+      "loss": 0.1777,
+      "step": 4487
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.768632340288571e-05,
+      "loss": 0.0811,
+      "step": 4488
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7681429647648584e-05,
+      "loss": 0.2242,
+      "step": 4489
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7676535597562644e-05,
+      "loss": 0.2497,
+      "step": 4490
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.767164125316604e-05,
+      "loss": 0.1397,
+      "step": 4491
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.766674661499694e-05,
+      "loss": 0.2897,
+      "step": 4492
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.766185168359357e-05,
+      "loss": 0.2474,
+      "step": 4493
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.765695645949417e-05,
+      "loss": 0.13,
+      "step": 4494
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7652060943237012e-05,
+      "loss": 0.1366,
+      "step": 4495
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7647165135360412e-05,
+      "loss": 0.1849,
+      "step": 4496
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7642269036402702e-05,
+      "loss": 0.2648,
+      "step": 4497
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.763737264690226e-05,
+      "loss": 0.1678,
+      "step": 4498
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7632475967397497e-05,
+      "loss": 0.1061,
+      "step": 4499
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7627578998426836e-05,
+      "loss": 0.2241,
+      "step": 4500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7622681740528754e-05,
+      "loss": 0.194,
+      "step": 4501
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7617784194241754e-05,
+      "loss": 0.2046,
+      "step": 4502
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7612886360104362e-05,
+      "loss": 0.1669,
+      "step": 4503
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7607988238655143e-05,
+      "loss": 0.1558,
+      "step": 4504
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7603089830432698e-05,
+      "loss": 0.2076,
+      "step": 4505
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.759819113597564e-05,
+      "loss": 0.2431,
+      "step": 4506
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7593292155822645e-05,
+      "loss": 0.2134,
+      "step": 4507
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7588392890512393e-05,
+      "loss": 0.0923,
+      "step": 4508
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.75834933405836e-05,
+      "loss": 0.1687,
+      "step": 4509
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.757859350657503e-05,
+      "loss": 0.1963,
+      "step": 4510
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.757369338902546e-05,
+      "loss": 0.1364,
+      "step": 4511
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7568792988473702e-05,
+      "loss": 0.2004,
+      "step": 4512
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.756389230545861e-05,
+      "loss": 0.1849,
+      "step": 4513
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7558991340519054e-05,
+      "loss": 0.1801,
+      "step": 4514
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7554090094193944e-05,
+      "loss": 0.1663,
+      "step": 4515
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7549188567022224e-05,
+      "loss": 0.1975,
+      "step": 4516
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7544286759542856e-05,
+      "loss": 0.1196,
+      "step": 4517
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7539384672294837e-05,
+      "loss": 0.1731,
+      "step": 4518
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.753448230581721e-05,
+      "loss": 0.1052,
+      "step": 4519
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7529579660649034e-05,
+      "loss": 0.1266,
+      "step": 4520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.75246767373294e-05,
+      "loss": 0.2189,
+      "step": 4521
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.751977353639743e-05,
+      "loss": 0.2432,
+      "step": 4522
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7514870058392275e-05,
+      "loss": 0.2633,
+      "step": 4523
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.750996630385313e-05,
+      "loss": 0.2537,
+      "step": 4524
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7505062273319198e-05,
+      "loss": 0.2081,
+      "step": 4525
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7500157967329733e-05,
+      "loss": 0.173,
+      "step": 4526
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7495253386424003e-05,
+      "loss": 0.2063,
+      "step": 4527
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7490348531141312e-05,
+      "loss": 0.2405,
+      "step": 4528
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7485443402021005e-05,
+      "loss": 0.19,
+      "step": 4529
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7480537999602444e-05,
+      "loss": 0.2552,
+      "step": 4530
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.747563232442502e-05,
+      "loss": 0.1361,
+      "step": 4531
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7470726377028168e-05,
+      "loss": 0.1953,
+      "step": 4532
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.746582015795134e-05,
+      "loss": 0.1432,
+      "step": 4533
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7460913667734012e-05,
+      "loss": 0.2228,
+      "step": 4534
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7456006906915717e-05,
+      "loss": 0.1245,
+      "step": 4535
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7451099876035986e-05,
+      "loss": 0.1576,
+      "step": 4536
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7446192575634396e-05,
+      "loss": 0.1395,
+      "step": 4537
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7441285006250565e-05,
+      "loss": 0.2504,
+      "step": 4538
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7436377168424104e-05,
+      "loss": 0.1729,
+      "step": 4539
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7431469062694698e-05,
+      "loss": 0.1776,
+      "step": 4540
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7426560689602025e-05,
+      "loss": 0.1645,
+      "step": 4541
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7421652049685823e-05,
+      "loss": 0.1977,
+      "step": 4542
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7416743143485826e-05,
+      "loss": 0.2042,
+      "step": 4543
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7411833971541832e-05,
+      "loss": 0.2266,
+      "step": 4544
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.740692453439364e-05,
+      "loss": 0.2686,
+      "step": 4545
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.740201483258109e-05,
+      "loss": 0.2285,
+      "step": 4546
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7397104866644058e-05,
+      "loss": 0.1496,
+      "step": 4547
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7392194637122434e-05,
+      "loss": 0.2205,
+      "step": 4548
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7387284144556152e-05,
+      "loss": 0.1278,
+      "step": 4549
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7382373389485154e-05,
+      "loss": 0.1311,
+      "step": 4550
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7377462372449447e-05,
+      "loss": 0.2036,
+      "step": 4551
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7372551093989022e-05,
+      "loss": 0.1588,
+      "step": 4552
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7367639554643936e-05,
+      "loss": 0.1497,
+      "step": 4553
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7362727754954248e-05,
+      "loss": 0.1925,
+      "step": 4554
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.735781569546007e-05,
+      "loss": 0.2468,
+      "step": 4555
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7352903376701518e-05,
+      "loss": 0.1937,
+      "step": 4556
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7347990799218757e-05,
+      "loss": 0.2331,
+      "step": 4557
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7343077963551965e-05,
+      "loss": 0.2078,
+      "step": 4558
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7338164870241363e-05,
+      "loss": 0.196,
+      "step": 4559
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7333251519827184e-05,
+      "loss": 0.2422,
+      "step": 4560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7328337912849707e-05,
+      "loss": 0.2114,
+      "step": 4561
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.732342404984922e-05,
+      "loss": 0.1979,
+      "step": 4562
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7318509931366056e-05,
+      "loss": 0.2108,
+      "step": 4563
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7313595557940568e-05,
+      "loss": 0.2094,
+      "step": 4564
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7308680930113138e-05,
+      "loss": 0.1803,
+      "step": 4565
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7303766048424178e-05,
+      "loss": 0.2062,
+      "step": 4566
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.729885091341412e-05,
+      "loss": 0.1465,
+      "step": 4567
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7293935525623435e-05,
+      "loss": 0.2027,
+      "step": 4568
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.728901988559262e-05,
+      "loss": 0.1202,
+      "step": 4569
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.728410399386219e-05,
+      "loss": 0.1746,
+      "step": 4570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.72791878509727e-05,
+      "loss": 0.1433,
+      "step": 4571
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7274271457464723e-05,
+      "loss": 0.2155,
+      "step": 4572
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.726935481387886e-05,
+      "loss": 0.1573,
+      "step": 4573
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7264437920755756e-05,
+      "loss": 0.205,
+      "step": 4574
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7259520778636058e-05,
+      "loss": 0.1336,
+      "step": 4575
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.725460338806045e-05,
+      "loss": 0.2347,
+      "step": 4576
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7249685749569664e-05,
+      "loss": 0.1901,
+      "step": 4577
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7244767863704424e-05,
+      "loss": 0.2995,
+      "step": 4578
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7239849731005505e-05,
+      "loss": 0.2335,
+      "step": 4579
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.72349313520137e-05,
+      "loss": 0.0949,
+      "step": 4580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7230012727269836e-05,
+      "loss": 0.226,
+      "step": 4581
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.722509385731476e-05,
+      "loss": 0.1539,
+      "step": 4582
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7220174742689354e-05,
+      "loss": 0.1797,
+      "step": 4583
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7215255383934517e-05,
+      "loss": 0.2311,
+      "step": 4584
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7210335781591174e-05,
+      "loss": 0.1191,
+      "step": 4585
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7205415936200296e-05,
+      "loss": 0.1578,
+      "step": 4586
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.720049584830285e-05,
+      "loss": 0.273,
+      "step": 4587
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7195575518439868e-05,
+      "loss": 0.1228,
+      "step": 4588
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7190654947152366e-05,
+      "loss": 0.2485,
+      "step": 4589
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7185734134981416e-05,
+      "loss": 0.231,
+      "step": 4590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7180813082468116e-05,
+      "loss": 0.2495,
+      "step": 4591
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.717589179015357e-05,
+      "loss": 0.2788,
+      "step": 4592
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7170970258578925e-05,
+      "loss": 0.1768,
+      "step": 4593
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7166048488285362e-05,
+      "loss": 0.1642,
+      "step": 4594
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7161126479814056e-05,
+      "loss": 0.1526,
+      "step": 4595
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.715620423370625e-05,
+      "loss": 0.192,
+      "step": 4596
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.715128175050318e-05,
+      "loss": 0.0951,
+      "step": 4597
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.714635903074612e-05,
+      "loss": 0.2669,
+      "step": 4598
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.714143607497637e-05,
+      "loss": 0.2022,
+      "step": 4599
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7136512883735265e-05,
+      "loss": 0.2085,
+      "step": 4600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7131589457564142e-05,
+      "loss": 0.1288,
+      "step": 4601
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7126665797004394e-05,
+      "loss": 0.1531,
+      "step": 4602
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7121741902597415e-05,
+      "loss": 0.1752,
+      "step": 4603
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7116817774884637e-05,
+      "loss": 0.121,
+      "step": 4604
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7111893414407514e-05,
+      "loss": 0.2095,
+      "step": 4605
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.710696882170753e-05,
+      "loss": 0.1764,
+      "step": 4606
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.710204399732618e-05,
+      "loss": 0.1242,
+      "step": 4607
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7097118941805016e-05,
+      "loss": 0.1592,
+      "step": 4608
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7092193655685575e-05,
+      "loss": 0.1726,
+      "step": 4609
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7087268139509448e-05,
+      "loss": 0.179,
+      "step": 4610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.708234239381824e-05,
+      "loss": 0.2146,
+      "step": 4611
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7077416419153584e-05,
+      "loss": 0.1788,
+      "step": 4612
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7072490216057148e-05,
+      "loss": 0.1526,
+      "step": 4613
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7067563785070598e-05,
+      "loss": 0.1882,
+      "step": 4614
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7062637126735654e-05,
+      "loss": 0.1996,
+      "step": 4615
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7057710241594046e-05,
+      "loss": 0.1372,
+      "step": 4616
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.705278313018753e-05,
+      "loss": 0.1835,
+      "step": 4617
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7047855793057894e-05,
+      "loss": 0.1344,
+      "step": 4618
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7042928230746942e-05,
+      "loss": 0.2435,
+      "step": 4619
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.70380004437965e-05,
+      "loss": 0.1805,
+      "step": 4620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7033072432748443e-05,
+      "loss": 0.2636,
+      "step": 4621
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7028144198144636e-05,
+      "loss": 0.2217,
+      "step": 4622
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7023215740526995e-05,
+      "loss": 0.2307,
+      "step": 4623
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7018287060437444e-05,
+      "loss": 0.1225,
+      "step": 4624
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.701335815841795e-05,
+      "loss": 0.1655,
+      "step": 4625
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7008429035010474e-05,
+      "loss": 0.176,
+      "step": 4626
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.700349969075704e-05,
+      "loss": 0.2029,
+      "step": 4627
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6998570126199664e-05,
+      "loss": 0.1436,
+      "step": 4628
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6993640341880403e-05,
+      "loss": 0.1262,
+      "step": 4629
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.698871033834134e-05,
+      "loss": 0.1284,
+      "step": 4630
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6983780116124565e-05,
+      "loss": 0.1492,
+      "step": 4631
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6978849675772204e-05,
+      "loss": 0.1993,
+      "step": 4632
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6973919017826417e-05,
+      "loss": 0.2247,
+      "step": 4633
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6968988142829362e-05,
+      "loss": 0.16,
+      "step": 4634
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6964057051323246e-05,
+      "loss": 0.2788,
+      "step": 4635
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.695912574385029e-05,
+      "loss": 0.2173,
+      "step": 4636
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6954194220952737e-05,
+      "loss": 0.2709,
+      "step": 4637
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6949262483172848e-05,
+      "loss": 0.1013,
+      "step": 4638
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6944330531052922e-05,
+      "loss": 0.1637,
+      "step": 4639
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6939398365135274e-05,
+      "loss": 0.1518,
+      "step": 4640
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.693446598596224e-05,
+      "loss": 0.136,
+      "step": 4641
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6929533394076185e-05,
+      "loss": 0.2148,
+      "step": 4642
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6924600590019492e-05,
+      "loss": 0.1748,
+      "step": 4643
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6919667574334576e-05,
+      "loss": 0.1224,
+      "step": 4644
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6914734347563855e-05,
+      "loss": 0.1872,
+      "step": 4645
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6909800910249796e-05,
+      "loss": 0.2604,
+      "step": 4646
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.690486726293488e-05,
+      "loss": 0.2216,
+      "step": 4647
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6899933406161603e-05,
+      "loss": 0.2477,
+      "step": 4648
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6894999340472493e-05,
+      "loss": 0.1629,
+      "step": 4649
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.689006506641009e-05,
+      "loss": 0.0982,
+      "step": 4650
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6885130584516973e-05,
+      "loss": 0.1876,
+      "step": 4651
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6880195895335737e-05,
+      "loss": 0.2224,
+      "step": 4652
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6875260999408982e-05,
+      "loss": 0.1572,
+      "step": 4653
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6870325897279367e-05,
+      "loss": 0.155,
+      "step": 4654
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6865390589489543e-05,
+      "loss": 0.143,
+      "step": 4655
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.68604550765822e-05,
+      "loss": 0.1619,
+      "step": 4656
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6855519359100038e-05,
+      "loss": 0.1379,
+      "step": 4657
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.685058343758579e-05,
+      "loss": 0.1733,
+      "step": 4658
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6845647312582207e-05,
+      "loss": 0.1612,
+      "step": 4659
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6840710984632064e-05,
+      "loss": 0.1538,
+      "step": 4660
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.683577445427816e-05,
+      "loss": 0.1766,
+      "step": 4661
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6830837722063308e-05,
+      "loss": 0.17,
+      "step": 4662
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.682590078853035e-05,
+      "loss": 0.1975,
+      "step": 4663
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6820963654222152e-05,
+      "loss": 0.2064,
+      "step": 4664
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6816026319681595e-05,
+      "loss": 0.2632,
+      "step": 4665
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6811088785451595e-05,
+      "loss": 0.2164,
+      "step": 4666
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.680615105207507e-05,
+      "loss": 0.1605,
+      "step": 4667
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6801213120094977e-05,
+      "loss": 0.1722,
+      "step": 4668
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.679627499005429e-05,
+      "loss": 0.0837,
+      "step": 4669
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6791336662495993e-05,
+      "loss": 0.0948,
+      "step": 4670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6786398137963117e-05,
+      "loss": 0.3047,
+      "step": 4671
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6781459416998694e-05,
+      "loss": 0.2049,
+      "step": 4672
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6776520500145785e-05,
+      "loss": 0.1878,
+      "step": 4673
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6771581387947468e-05,
+      "loss": 0.189,
+      "step": 4674
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6766642080946844e-05,
+      "loss": 0.1988,
+      "step": 4675
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.676170257968704e-05,
+      "loss": 0.1624,
+      "step": 4676
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6756762884711203e-05,
+      "loss": 0.2269,
+      "step": 4677
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6751822996562497e-05,
+      "loss": 0.2456,
+      "step": 4678
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6746882915784112e-05,
+      "loss": 0.2172,
+      "step": 4679
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6741942642919252e-05,
+      "loss": 0.1772,
+      "step": 4680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6737002178511158e-05,
+      "loss": 0.1406,
+      "step": 4681
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6732061523103068e-05,
+      "loss": 0.253,
+      "step": 4682
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6727120677238263e-05,
+      "loss": 0.2155,
+      "step": 4683
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6722179641460033e-05,
+      "loss": 0.1537,
+      "step": 4684
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.671723841631169e-05,
+      "loss": 0.1348,
+      "step": 4685
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.671229700233658e-05,
+      "loss": 0.1329,
+      "step": 4686
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6707355400078048e-05,
+      "loss": 0.1271,
+      "step": 4687
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6702413610079468e-05,
+      "loss": 0.2126,
+      "step": 4688
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6697471632884242e-05,
+      "loss": 0.1356,
+      "step": 4689
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6692529469035793e-05,
+      "loss": 0.1332,
+      "step": 4690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6687587119077552e-05,
+      "loss": 0.2231,
+      "step": 4691
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6682644583552973e-05,
+      "loss": 0.1291,
+      "step": 4692
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.667770186300555e-05,
+      "loss": 0.1927,
+      "step": 4693
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6672758957978772e-05,
+      "loss": 0.296,
+      "step": 4694
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.666781586901616e-05,
+      "loss": 0.2052,
+      "step": 4695
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6662872596661253e-05,
+      "loss": 0.1839,
+      "step": 4696
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6657929141457614e-05,
+      "loss": 0.1748,
+      "step": 4697
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6652985503948815e-05,
+      "loss": 0.2035,
+      "step": 4698
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6648041684678472e-05,
+      "loss": 0.1879,
+      "step": 4699
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.664309768419019e-05,
+      "loss": 0.1704,
+      "step": 4700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6638153503027613e-05,
+      "loss": 0.1195,
+      "step": 4701
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6633209141734413e-05,
+      "loss": 0.1076,
+      "step": 4702
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.662826460085425e-05,
+      "loss": 0.238,
+      "step": 4703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6623319880930836e-05,
+      "loss": 0.1559,
+      "step": 4704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6618374982507887e-05,
+      "loss": 0.1532,
+      "step": 4705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.661342990612914e-05,
+      "loss": 0.1469,
+      "step": 4706
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.660848465233836e-05,
+      "loss": 0.165,
+      "step": 4707
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.660353922167932e-05,
+      "loss": 0.1605,
+      "step": 4708
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6598593614695813e-05,
+      "loss": 0.1296,
+      "step": 4709
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6593647831931666e-05,
+      "loss": 0.1017,
+      "step": 4710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6588701873930708e-05,
+      "loss": 0.1768,
+      "step": 4711
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6583755741236796e-05,
+      "loss": 0.1879,
+      "step": 4712
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.657880943439381e-05,
+      "loss": 0.2469,
+      "step": 4713
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6573862953945632e-05,
+      "loss": 0.1985,
+      "step": 4714
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6568916300436176e-05,
+      "loss": 0.2028,
+      "step": 4715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6563969474409392e-05,
+      "loss": 0.1581,
+      "step": 4716
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6559022476409208e-05,
+      "loss": 0.148,
+      "step": 4717
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6554075306979607e-05,
+      "loss": 0.1895,
+      "step": 4718
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6549127966664575e-05,
+      "loss": 0.1475,
+      "step": 4719
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6544180456008117e-05,
+      "loss": 0.2276,
+      "step": 4720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6539232775554263e-05,
+      "loss": 0.2096,
+      "step": 4721
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.653428492584705e-05,
+      "loss": 0.2197,
+      "step": 4722
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.652933690743055e-05,
+      "loss": 0.1911,
+      "step": 4723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6524388720848837e-05,
+      "loss": 0.1622,
+      "step": 4724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.651944036664602e-05,
+      "loss": 0.1938,
+      "step": 4725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6514491845366218e-05,
+      "loss": 0.2035,
+      "step": 4726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.650954315755356e-05,
+      "loss": 0.2292,
+      "step": 4727
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6504594303752206e-05,
+      "loss": 0.1862,
+      "step": 4728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6499645284506322e-05,
+      "loss": 0.134,
+      "step": 4729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6494696100360115e-05,
+      "loss": 0.2358,
+      "step": 4730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6489746751857786e-05,
+      "loss": 0.0978,
+      "step": 4731
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6484797239543564e-05,
+      "loss": 0.1688,
+      "step": 4732
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.64798475639617e-05,
+      "loss": 0.1471,
+      "step": 4733
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6474897725656445e-05,
+      "loss": 0.14,
+      "step": 4734
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6469947725172097e-05,
+      "loss": 0.2108,
+      "step": 4735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6464997563052945e-05,
+      "loss": 0.2442,
+      "step": 4736
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.646004723984331e-05,
+      "loss": 0.1847,
+      "step": 4737
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6455096756087528e-05,
+      "loss": 0.1833,
+      "step": 4738
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6450146112329955e-05,
+      "loss": 0.151,
+      "step": 4739
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6445195309114953e-05,
+      "loss": 0.175,
+      "step": 4740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.644024434698692e-05,
+      "loss": 0.2314,
+      "step": 4741
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6435293226490253e-05,
+      "loss": 0.13,
+      "step": 4742
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6430341948169382e-05,
+      "loss": 0.1751,
+      "step": 4743
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6425390512568747e-05,
+      "loss": 0.1745,
+      "step": 4744
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6420438920232796e-05,
+      "loss": 0.1709,
+      "step": 4745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6415487171706016e-05,
+      "loss": 0.0882,
+      "step": 4746
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6410535267532894e-05,
+      "loss": 0.1876,
+      "step": 4747
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6405583208257935e-05,
+      "loss": 0.1593,
+      "step": 4748
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.640063099442567e-05,
+      "loss": 0.2782,
+      "step": 4749
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.639567862658065e-05,
+      "loss": 0.1637,
+      "step": 4750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6390726105267415e-05,
+      "loss": 0.1711,
+      "step": 4751
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.638577343103056e-05,
+      "loss": 0.2625,
+      "step": 4752
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6380820604414673e-05,
+      "loss": 0.1418,
+      "step": 4753
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.637586762596436e-05,
+      "loss": 0.1094,
+      "step": 4754
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6370914496224258e-05,
+      "loss": 0.188,
+      "step": 4755
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6365961215739002e-05,
+      "loss": 0.1468,
+      "step": 4756
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.636100778505326e-05,
+      "loss": 0.1905,
+      "step": 4757
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6356054204711704e-05,
+      "loss": 0.1728,
+      "step": 4758
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6351100475259026e-05,
+      "loss": 0.1659,
+      "step": 4759
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.634614659723994e-05,
+      "loss": 0.1713,
+      "step": 4760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.634119257119917e-05,
+      "loss": 0.2219,
+      "step": 4761
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6336238397681462e-05,
+      "loss": 0.2031,
+      "step": 4762
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.633128407723157e-05,
+      "loss": 0.2332,
+      "step": 4763
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6326329610394272e-05,
+      "loss": 0.1305,
+      "step": 4764
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6321374997714357e-05,
+      "loss": 0.1114,
+      "step": 4765
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6316420239736636e-05,
+      "loss": 0.2304,
+      "step": 4766
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.631146533700592e-05,
+      "loss": 0.1593,
+      "step": 4767
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6306510290067064e-05,
+      "loss": 0.0964,
+      "step": 4768
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6301555099464914e-05,
+      "loss": 0.1744,
+      "step": 4769
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6296599765744344e-05,
+      "loss": 0.0982,
+      "step": 4770
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.629164428945024e-05,
+      "loss": 0.1681,
+      "step": 4771
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6286688671127496e-05,
+      "loss": 0.2308,
+      "step": 4772
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6281732911321038e-05,
+      "loss": 0.1831,
+      "step": 4773
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.62767770105758e-05,
+      "loss": 0.1466,
+      "step": 4774
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6271820969436727e-05,
+      "loss": 0.1819,
+      "step": 4775
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.626686478844878e-05,
+      "loss": 0.1643,
+      "step": 4776
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.626190846815695e-05,
+      "loss": 0.2218,
+      "step": 4777
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.625695200910622e-05,
+      "loss": 0.097,
+      "step": 4778
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.62519954118416e-05,
+      "loss": 0.1393,
+      "step": 4779
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6247038676908127e-05,
+      "loss": 0.1628,
+      "step": 4780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.624208180485083e-05,
+      "loss": 0.1032,
+      "step": 4781
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6237124796214765e-05,
+      "loss": 0.1368,
+      "step": 4782
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.623216765154501e-05,
+      "loss": 0.1294,
+      "step": 4783
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.622721037138665e-05,
+      "loss": 0.1938,
+      "step": 4784
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6222252956284776e-05,
+      "loss": 0.1539,
+      "step": 4785
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.621729540678451e-05,
+      "loss": 0.1439,
+      "step": 4786
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6212337723430977e-05,
+      "loss": 0.0873,
+      "step": 4787
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6207379906769332e-05,
+      "loss": 0.2039,
+      "step": 4788
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6202421957344725e-05,
+      "loss": 0.1952,
+      "step": 4789
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6197463875702327e-05,
+      "loss": 0.13,
+      "step": 4790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.619250566238734e-05,
+      "loss": 0.2075,
+      "step": 4791
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6187547317944953e-05,
+      "loss": 0.1205,
+      "step": 4792
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.618258884292039e-05,
+      "loss": 0.161,
+      "step": 4793
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6177630237858884e-05,
+      "loss": 0.1696,
+      "step": 4794
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.617267150330568e-05,
+      "loss": 0.1314,
+      "step": 4795
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6167712639806036e-05,
+      "loss": 0.1219,
+      "step": 4796
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6162753647905227e-05,
+      "loss": 0.1323,
+      "step": 4797
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.615779452814854e-05,
+      "loss": 0.1104,
+      "step": 4798
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6152835281081288e-05,
+      "loss": 0.1436,
+      "step": 4799
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6147875907248774e-05,
+      "loss": 0.1685,
+      "step": 4800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6142916407196332e-05,
+      "loss": 0.1048,
+      "step": 4801
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.613795678146931e-05,
+      "loss": 0.2161,
+      "step": 4802
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6132997030613066e-05,
+      "loss": 0.0942,
+      "step": 4803
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6128037155172966e-05,
+      "loss": 0.1476,
+      "step": 4804
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.612307715569441e-05,
+      "loss": 0.2059,
+      "step": 4805
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.611811703272278e-05,
+      "loss": 0.1205,
+      "step": 4806
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6113156786803497e-05,
+      "loss": 0.1647,
+      "step": 4807
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.610819641848199e-05,
+      "loss": 0.1933,
+      "step": 4808
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6103235928303696e-05,
+      "loss": 0.1893,
+      "step": 4809
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6098275316814066e-05,
+      "loss": 0.162,
+      "step": 4810
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6093314584558574e-05,
+      "loss": 0.2604,
+      "step": 4811
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6088353732082687e-05,
+      "loss": 0.1633,
+      "step": 4812
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.608339275993191e-05,
+      "loss": 0.1844,
+      "step": 4813
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6078431668651745e-05,
+      "loss": 0.1663,
+      "step": 4814
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.607347045878771e-05,
+      "loss": 0.141,
+      "step": 4815
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6068509130885345e-05,
+      "loss": 0.2393,
+      "step": 4816
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6063547685490186e-05,
+      "loss": 0.198,
+      "step": 4817
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.605858612314779e-05,
+      "loss": 0.1457,
+      "step": 4818
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6053624444403744e-05,
+      "loss": 0.1747,
+      "step": 4819
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.604866264980361e-05,
+      "loss": 0.1744,
+      "step": 4820
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6043700739893e-05,
+      "loss": 0.1089,
+      "step": 4821
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.603873871521752e-05,
+      "loss": 0.1829,
+      "step": 4822
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.603377657632279e-05,
+      "loss": 0.0785,
+      "step": 4823
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.602881432375445e-05,
+      "loss": 0.2077,
+      "step": 4824
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.602385195805814e-05,
+      "loss": 0.09,
+      "step": 4825
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.601888947977952e-05,
+      "loss": 0.2265,
+      "step": 4826
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6013926889464267e-05,
+      "loss": 0.1716,
+      "step": 4827
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6008964187658063e-05,
+      "loss": 0.126,
+      "step": 4828
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6004001374906602e-05,
+      "loss": 0.1548,
+      "step": 4829
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.59990384517556e-05,
+      "loss": 0.1861,
+      "step": 4830
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5994075418750766e-05,
+      "loss": 0.2166,
+      "step": 4831
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.598911227643784e-05,
+      "loss": 0.2495,
+      "step": 4832
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.598414902536257e-05,
+      "loss": 0.1813,
+      "step": 4833
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.597918566607071e-05,
+      "loss": 0.1832,
+      "step": 4834
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.597422219910803e-05,
+      "loss": 0.2017,
+      "step": 4835
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5969258625020304e-05,
+      "loss": 0.203,
+      "step": 4836
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.596429494435333e-05,
+      "loss": 0.1775,
+      "step": 4837
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5959331157652916e-05,
+      "loss": 0.2042,
+      "step": 4838
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5954367265464863e-05,
+      "loss": 0.1792,
+      "step": 4839
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5949403268335015e-05,
+      "loss": 0.1411,
+      "step": 4840
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5944439166809206e-05,
+      "loss": 0.1727,
+      "step": 4841
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.593947496143328e-05,
+      "loss": 0.1811,
+      "step": 4842
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5934510652753105e-05,
+      "loss": 0.1509,
+      "step": 4843
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.592954624131456e-05,
+      "loss": 0.2033,
+      "step": 4844
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5924581727663513e-05,
+      "loss": 0.1971,
+      "step": 4845
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.591961711234587e-05,
+      "loss": 0.1804,
+      "step": 4846
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.591465239590754e-05,
+      "loss": 0.1412,
+      "step": 4847
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5909687578894438e-05,
+      "loss": 0.1902,
+      "step": 4848
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5904722661852494e-05,
+      "loss": 0.2786,
+      "step": 4849
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5899757645327645e-05,
+      "loss": 0.2147,
+      "step": 4850
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.589479252986584e-05,
+      "loss": 0.1138,
+      "step": 4851
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5889827316013053e-05,
+      "loss": 0.246,
+      "step": 4852
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5884862004315244e-05,
+      "loss": 0.1917,
+      "step": 4853
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.58798965953184e-05,
+      "loss": 0.1398,
+      "step": 4854
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5874931089568524e-05,
+      "loss": 0.1978,
+      "step": 4855
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5869965487611606e-05,
+      "loss": 0.1906,
+      "step": 4856
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.586499978999367e-05,
+      "loss": 0.1054,
+      "step": 4857
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5860033997260743e-05,
+      "loss": 0.1292,
+      "step": 4858
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5855068109958855e-05,
+      "loss": 0.1576,
+      "step": 4859
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.585010212863406e-05,
+      "loss": 0.2018,
+      "step": 4860
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.584513605383241e-05,
+      "loss": 0.2255,
+      "step": 4861
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5840169886099973e-05,
+      "loss": 0.274,
+      "step": 4862
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.583520362598283e-05,
+      "loss": 0.1784,
+      "step": 4863
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5830237274027065e-05,
+      "loss": 0.1443,
+      "step": 4864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5825270830778775e-05,
+      "loss": 0.1485,
+      "step": 4865
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5820304296784078e-05,
+      "loss": 0.1354,
+      "step": 4866
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.581533767258908e-05,
+      "loss": 0.1378,
+      "step": 4867
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5810370958739908e-05,
+      "loss": 0.15,
+      "step": 4868
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5805404155782712e-05,
+      "loss": 0.2078,
+      "step": 4869
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5800437264263627e-05,
+      "loss": 0.1742,
+      "step": 4870
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5795470284728817e-05,
+      "loss": 0.1315,
+      "step": 4871
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5790503217724448e-05,
+      "loss": 0.1919,
+      "step": 4872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5785536063796694e-05,
+      "loss": 0.2014,
+      "step": 4873
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5780568823491742e-05,
+      "loss": 0.1545,
+      "step": 4874
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5775601497355793e-05,
+      "loss": 0.1876,
+      "step": 4875
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5770634085935037e-05,
+      "loss": 0.2141,
+      "step": 4876
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.576566658977571e-05,
+      "loss": 0.1588,
+      "step": 4877
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5760699009424018e-05,
+      "loss": 0.1212,
+      "step": 4878
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.57557313454262e-05,
+      "loss": 0.1379,
+      "step": 4879
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.57507635983285e-05,
+      "loss": 0.1351,
+      "step": 4880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5745795768677162e-05,
+      "loss": 0.1654,
+      "step": 4881
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5740827857018457e-05,
+      "loss": 0.1616,
+      "step": 4882
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5735859863898648e-05,
+      "loss": 0.235,
+      "step": 4883
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5730891789864018e-05,
+      "loss": 0.2604,
+      "step": 4884
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.572592363546085e-05,
+      "loss": 0.1045,
+      "step": 4885
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5720955401235436e-05,
+      "loss": 0.0975,
+      "step": 4886
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5715987087734085e-05,
+      "loss": 0.1252,
+      "step": 4887
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5711018695503122e-05,
+      "loss": 0.1246,
+      "step": 4888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5706050225088847e-05,
+      "loss": 0.1507,
+      "step": 4889
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5701081677037608e-05,
+      "loss": 0.1738,
+      "step": 4890
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5696113051895736e-05,
+      "loss": 0.0803,
+      "step": 4891
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.569114435020958e-05,
+      "loss": 0.1966,
+      "step": 4892
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5686175572525497e-05,
+      "loss": 0.1223,
+      "step": 4893
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5681206719389857e-05,
+      "loss": 0.1613,
+      "step": 4894
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5676237791349024e-05,
+      "loss": 0.0981,
+      "step": 4895
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5671268788949387e-05,
+      "loss": 0.1022,
+      "step": 4896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5666299712737327e-05,
+      "loss": 0.0806,
+      "step": 4897
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.566133056325925e-05,
+      "loss": 0.0787,
+      "step": 4898
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5656361341061558e-05,
+      "loss": 0.1199,
+      "step": 4899
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5651392046690664e-05,
+      "loss": 0.0632,
+      "step": 4900
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5646422680692983e-05,
+      "loss": 0.1288,
+      "step": 4901
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5641453243614957e-05,
+      "loss": 0.1067,
+      "step": 4902
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.563648373600301e-05,
+      "loss": 0.1068,
+      "step": 4903
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.56315141584036e-05,
+      "loss": 0.0929,
+      "step": 4904
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.562654451136317e-05,
+      "loss": 0.083,
+      "step": 4905
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5621574795428185e-05,
+      "loss": 0.1403,
+      "step": 4906
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.561660501114511e-05,
+      "loss": 0.104,
+      "step": 4907
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5611635159060425e-05,
+      "loss": 0.1343,
+      "step": 4908
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.56066652397206e-05,
+      "loss": 0.1193,
+      "step": 4909
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5601695253672137e-05,
+      "loss": 0.1676,
+      "step": 4910
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5596725201461536e-05,
+      "loss": 0.1187,
+      "step": 4911
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.559175508363529e-05,
+      "loss": 0.1064,
+      "step": 4912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5586784900739926e-05,
+      "loss": 0.1044,
+      "step": 4913
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5581814653321946e-05,
+      "loss": 0.1291,
+      "step": 4914
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5576844341927885e-05,
+      "loss": 0.0895,
+      "step": 4915
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.557187396710428e-05,
+      "loss": 0.1009,
+      "step": 4916
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5566903529397665e-05,
+      "loss": 0.0721,
+      "step": 4917
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5561933029354585e-05,
+      "loss": 0.0974,
+      "step": 4918
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.555696246752161e-05,
+      "loss": 0.18,
+      "step": 4919
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5551991844445277e-05,
+      "loss": 0.081,
+      "step": 4920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.554702116067217e-05,
+      "loss": 0.1201,
+      "step": 4921
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.554205041674886e-05,
+      "loss": 0.121,
+      "step": 4922
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5537079613221926e-05,
+      "loss": 0.0935,
+      "step": 4923
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5532108750637957e-05,
+      "loss": 0.1059,
+      "step": 4924
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.552713782954354e-05,
+      "loss": 0.1498,
+      "step": 4925
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5522166850485283e-05,
+      "loss": 0.1545,
+      "step": 4926
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5517195814009796e-05,
+      "loss": 0.0991,
+      "step": 4927
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.551222472066368e-05,
+      "loss": 0.1261,
+      "step": 4928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5507253570993564e-05,
+      "loss": 0.0651,
+      "step": 4929
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.550228236554607e-05,
+      "loss": 0.0914,
+      "step": 4930
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549731110486783e-05,
+      "loss": 0.0974,
+      "step": 4931
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5492339789505475e-05,
+      "loss": 0.1038,
+      "step": 4932
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5487368420005667e-05,
+      "loss": 0.1002,
+      "step": 4933
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5482396996915033e-05,
+      "loss": 0.088,
+      "step": 4934
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5477425520780244e-05,
+      "loss": 0.1015,
+      "step": 4935
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5472453992147953e-05,
+      "loss": 0.1555,
+      "step": 4936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5467482411564833e-05,
+      "loss": 0.0809,
+      "step": 4937
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5462510779577558e-05,
+      "loss": 0.1441,
+      "step": 4938
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5457539096732798e-05,
+      "loss": 0.1707,
+      "step": 4939
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.545256736357724e-05,
+      "loss": 0.1297,
+      "step": 4940
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5447595580657576e-05,
+      "loss": 0.1277,
+      "step": 4941
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54426237485205e-05,
+      "loss": 0.1352,
+      "step": 4942
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5437651867712713e-05,
+      "loss": 0.0916,
+      "step": 4943
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5432679938780928e-05,
+      "loss": 0.0989,
+      "step": 4944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542770796227184e-05,
+      "loss": 0.0909,
+      "step": 4945
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5422735938732178e-05,
+      "loss": 0.1034,
+      "step": 4946
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.541776386870866e-05,
+      "loss": 0.148,
+      "step": 4947
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5412791752748014e-05,
+      "loss": 0.1666,
+      "step": 4948
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5407819591396964e-05,
+      "loss": 0.1404,
+      "step": 4949
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5402847385202256e-05,
+      "loss": 0.0424,
+      "step": 4950
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5397875134710624e-05,
+      "loss": 0.1468,
+      "step": 4951
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5392902840468827e-05,
+      "loss": 0.0778,
+      "step": 4952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5387930503023598e-05,
+      "loss": 0.141,
+      "step": 4953
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5382958122921706e-05,
+      "loss": 0.1348,
+      "step": 4954
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5377985700709913e-05,
+      "loss": 0.1198,
+      "step": 4955
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.537301323693498e-05,
+      "loss": 0.1074,
+      "step": 4956
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5368040732143667e-05,
+      "loss": 0.1135,
+      "step": 4957
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536306818688277e-05,
+      "loss": 0.1053,
+      "step": 4958
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5358095601699047e-05,
+      "loss": 0.0886,
+      "step": 4959
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5353122977139292e-05,
+      "loss": 0.1196,
+      "step": 4960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5348150313750297e-05,
+      "loss": 0.1032,
+      "step": 4961
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5343177612078845e-05,
+      "loss": 0.1792,
+      "step": 4962
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5338204872671733e-05,
+      "loss": 0.0637,
+      "step": 4963
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5333232096075764e-05,
+      "loss": 0.141,
+      "step": 4964
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5328259282837736e-05,
+      "loss": 0.1102,
+      "step": 4965
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5323286433504477e-05,
+      "loss": 0.0994,
+      "step": 4966
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.531831354862277e-05,
+      "loss": 0.0953,
+      "step": 4967
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5313340628739456e-05,
+      "loss": 0.0874,
+      "step": 4968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.530836767440134e-05,
+      "loss": 0.1155,
+      "step": 4969
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5303394686155255e-05,
+      "loss": 0.139,
+      "step": 4970
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.529842166454802e-05,
+      "loss": 0.1042,
+      "step": 4971
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5293448610126475e-05,
+      "loss": 0.1477,
+      "step": 4972
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.528847552343745e-05,
+      "loss": 0.1256,
+      "step": 4973
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5283502405027783e-05,
+      "loss": 0.1194,
+      "step": 4974
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.527852925544432e-05,
+      "loss": 0.1485,
+      "step": 4975
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5273556075233895e-05,
+      "loss": 0.1223,
+      "step": 4976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5268582864943376e-05,
+      "loss": 0.1105,
+      "step": 4977
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.52636096251196e-05,
+      "loss": 0.1529,
+      "step": 4978
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5258636356309425e-05,
+      "loss": 0.1193,
+      "step": 4979
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5253663059059714e-05,
+      "loss": 0.1552,
+      "step": 4980
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5248689733917327e-05,
+      "loss": 0.1427,
+      "step": 4981
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5243716381429117e-05,
+      "loss": 0.1346,
+      "step": 4982
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5238743002141977e-05,
+      "loss": 0.12,
+      "step": 4983
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.523376959660275e-05,
+      "loss": 0.1206,
+      "step": 4984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5228796165358326e-05,
+      "loss": 0.1513,
+      "step": 4985
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5223822708955577e-05,
+      "loss": 0.0769,
+      "step": 4986
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5218849227941382e-05,
+      "loss": 0.0971,
+      "step": 4987
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5213875722862622e-05,
+      "loss": 0.1022,
+      "step": 4988
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5208902194266185e-05,
+      "loss": 0.1185,
+      "step": 4989
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.520392864269895e-05,
+      "loss": 0.1002,
+      "step": 4990
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5198955068707821e-05,
+      "loss": 0.1507,
+      "step": 4991
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5193981472839673e-05,
+      "loss": 0.1161,
+      "step": 4992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.518900785564141e-05,
+      "loss": 0.0725,
+      "step": 4993
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5184034217659929e-05,
+      "loss": 0.1007,
+      "step": 4994
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5179060559442125e-05,
+      "loss": 0.0974,
+      "step": 4995
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5174086881534897e-05,
+      "loss": 0.0659,
+      "step": 4996
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5169113184485162e-05,
+      "loss": 0.1919,
+      "step": 4997
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.516413946883981e-05,
+      "loss": 0.0855,
+      "step": 4998
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5159165735145756e-05,
+      "loss": 0.0808,
+      "step": 4999
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5154191983949908e-05,
+      "loss": 0.1008,
+      "step": 5000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5149218215799178e-05,
+      "loss": 0.0553,
+      "step": 5001
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5144244431240483e-05,
+      "loss": 0.0988,
+      "step": 5002
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5139270630820729e-05,
+      "loss": 0.0885,
+      "step": 5003
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5134296815086838e-05,
+      "loss": 0.1777,
+      "step": 5004
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5129322984585732e-05,
+      "loss": 0.1009,
+      "step": 5005
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5124349139864326e-05,
+      "loss": 0.0936,
+      "step": 5006
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5119375281469541e-05,
+      "loss": 0.1158,
+      "step": 5007
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5114401409948312e-05,
+      "loss": 0.1348,
+      "step": 5008
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5109427525847546e-05,
+      "loss": 0.1038,
+      "step": 5009
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.510445362971418e-05,
+      "loss": 0.2331,
+      "step": 5010
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5099479722095141e-05,
+      "loss": 0.1144,
+      "step": 5011
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5094505803537354e-05,
+      "loss": 0.1369,
+      "step": 5012
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5089531874587752e-05,
+      "loss": 0.0937,
+      "step": 5013
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5084557935793264e-05,
+      "loss": 0.1253,
+      "step": 5014
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5079583987700822e-05,
+      "loss": 0.0909,
+      "step": 5015
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5074610030857365e-05,
+      "loss": 0.0635,
+      "step": 5016
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5069636065809816e-05,
+      "loss": 0.0732,
+      "step": 5017
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5064662093105119e-05,
+      "loss": 0.1266,
+      "step": 5018
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5059688113290206e-05,
+      "loss": 0.105,
+      "step": 5019
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5054714126912012e-05,
+      "loss": 0.14,
+      "step": 5020
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5049740134517479e-05,
+      "loss": 0.0744,
+      "step": 5021
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5044766136653547e-05,
+      "loss": 0.0857,
+      "step": 5022
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5039792133867143e-05,
+      "loss": 0.1779,
+      "step": 5023
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5034818126705219e-05,
+      "loss": 0.1361,
+      "step": 5024
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5029844115714707e-05,
+      "loss": 0.1046,
+      "step": 5025
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.502487010144255e-05,
+      "loss": 0.1424,
+      "step": 5026
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5019896084435687e-05,
+      "loss": 0.0684,
+      "step": 5027
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5014922065241062e-05,
+      "loss": 0.1269,
+      "step": 5028
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5009948044405609e-05,
+      "loss": 0.1191,
+      "step": 5029
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5004974022476278e-05,
+      "loss": 0.0697,
+      "step": 5030
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5e-05,
+      "loss": 0.1256,
+      "step": 5031
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4995025977523722e-05,
+      "loss": 0.1009,
+      "step": 5032
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4990051955594392e-05,
+      "loss": 0.1121,
+      "step": 5033
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4985077934758944e-05,
+      "loss": 0.1282,
+      "step": 5034
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4980103915564314e-05,
+      "loss": 0.1296,
+      "step": 5035
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4975129898557453e-05,
+      "loss": 0.1195,
+      "step": 5036
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4970155884285292e-05,
+      "loss": 0.1345,
+      "step": 5037
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4965181873294784e-05,
+      "loss": 0.1202,
+      "step": 5038
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.496020786613286e-05,
+      "loss": 0.1467,
+      "step": 5039
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4955233863346455e-05,
+      "loss": 0.0906,
+      "step": 5040
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.495025986548252e-05,
+      "loss": 0.0783,
+      "step": 5041
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4945285873087992e-05,
+      "loss": 0.1341,
+      "step": 5042
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4940311886709797e-05,
+      "loss": 0.0542,
+      "step": 5043
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4935337906894885e-05,
+      "loss": 0.1138,
+      "step": 5044
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4930363934190185e-05,
+      "loss": 0.1037,
+      "step": 5045
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4925389969142641e-05,
+      "loss": 0.1442,
+      "step": 5046
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4920416012299182e-05,
+      "loss": 0.1753,
+      "step": 5047
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4915442064206737e-05,
+      "loss": 0.0836,
+      "step": 5048
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.491046812541225e-05,
+      "loss": 0.1344,
+      "step": 5049
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4905494196462651e-05,
+      "loss": 0.0771,
+      "step": 5050
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4900520277904861e-05,
+      "loss": 0.1226,
+      "step": 5051
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4895546370285822e-05,
+      "loss": 0.1522,
+      "step": 5052
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4890572474152453e-05,
+      "loss": 0.0864,
+      "step": 5053
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4885598590051696e-05,
+      "loss": 0.1002,
+      "step": 5054
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.488062471853046e-05,
+      "loss": 0.1394,
+      "step": 5055
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4875650860135677e-05,
+      "loss": 0.1691,
+      "step": 5056
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4870677015414269e-05,
+      "loss": 0.1162,
+      "step": 5057
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4865703184913168e-05,
+      "loss": 0.1307,
+      "step": 5058
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4860729369179275e-05,
+      "loss": 0.0709,
+      "step": 5059
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4855755568759522e-05,
+      "loss": 0.1651,
+      "step": 5060
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4850781784200822e-05,
+      "loss": 0.0792,
+      "step": 5061
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.484580801605009e-05,
+      "loss": 0.1286,
+      "step": 5062
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4840834264854248e-05,
+      "loss": 0.0884,
+      "step": 5063
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4835860531160192e-05,
+      "loss": 0.0845,
+      "step": 5064
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.483088681551484e-05,
+      "loss": 0.0967,
+      "step": 5065
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4825913118465102e-05,
+      "loss": 0.1099,
+      "step": 5066
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4820939440557881e-05,
+      "loss": 0.1047,
+      "step": 5067
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4815965782340075e-05,
+      "loss": 0.0804,
+      "step": 5068
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4810992144358593e-05,
+      "loss": 0.1146,
+      "step": 5069
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4806018527160327e-05,
+      "loss": 0.0684,
+      "step": 5070
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4801044931292186e-05,
+      "loss": 0.0897,
+      "step": 5071
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4796071357301051e-05,
+      "loss": 0.0968,
+      "step": 5072
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4791097805733818e-05,
+      "loss": 0.1486,
+      "step": 5073
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4786124277137379e-05,
+      "loss": 0.116,
+      "step": 5074
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4781150772058622e-05,
+      "loss": 0.0551,
+      "step": 5075
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4776177291044425e-05,
+      "loss": 0.1108,
+      "step": 5076
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4771203834641677e-05,
+      "loss": 0.1405,
+      "step": 5077
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.476623040339725e-05,
+      "loss": 0.0531,
+      "step": 5078
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4761256997858031e-05,
+      "loss": 0.0937,
+      "step": 5079
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4756283618570883e-05,
+      "loss": 0.081,
+      "step": 5080
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4751310266082677e-05,
+      "loss": 0.0858,
+      "step": 5081
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4746336940940283e-05,
+      "loss": 0.1415,
+      "step": 5082
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4741363643690579e-05,
+      "loss": 0.085,
+      "step": 5083
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4736390374880402e-05,
+      "loss": 0.1077,
+      "step": 5084
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4731417135056624e-05,
+      "loss": 0.1242,
+      "step": 5085
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.47264439247661e-05,
+      "loss": 0.1241,
+      "step": 5086
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4721470744555687e-05,
+      "loss": 0.1062,
+      "step": 5087
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.471649759497222e-05,
+      "loss": 0.1624,
+      "step": 5088
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4711524476562554e-05,
+      "loss": 0.1407,
+      "step": 5089
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4706551389873526e-05,
+      "loss": 0.1764,
+      "step": 5090
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4701578335451983e-05,
+      "loss": 0.0726,
+      "step": 5091
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4696605313844753e-05,
+      "loss": 0.2059,
+      "step": 5092
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4691632325598661e-05,
+      "loss": 0.1539,
+      "step": 5093
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4686659371260549e-05,
+      "loss": 0.0901,
+      "step": 5094
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4681686451377227e-05,
+      "loss": 0.0933,
+      "step": 5095
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.467671356649553e-05,
+      "loss": 0.1264,
+      "step": 5096
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4671740717162263e-05,
+      "loss": 0.0983,
+      "step": 5097
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4666767903924238e-05,
+      "loss": 0.1522,
+      "step": 5098
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.466179512732827e-05,
+      "loss": 0.0671,
+      "step": 5099
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4656822387921163e-05,
+      "loss": 0.1174,
+      "step": 5100
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4651849686249707e-05,
+      "loss": 0.1172,
+      "step": 5101
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4646877022860707e-05,
+      "loss": 0.1182,
+      "step": 5102
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4641904398300952e-05,
+      "loss": 0.1268,
+      "step": 5103
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4636931813117237e-05,
+      "loss": 0.1412,
+      "step": 5104
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4631959267856335e-05,
+      "loss": 0.0942,
+      "step": 5105
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4626986763065027e-05,
+      "loss": 0.1453,
+      "step": 5106
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4622014299290088e-05,
+      "loss": 0.1087,
+      "step": 5107
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4617041877078298e-05,
+      "loss": 0.1224,
+      "step": 5108
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4612069496976406e-05,
+      "loss": 0.1396,
+      "step": 5109
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4607097159531178e-05,
+      "loss": 0.0978,
+      "step": 5110
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4602124865289376e-05,
+      "loss": 0.1506,
+      "step": 5111
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.459715261479775e-05,
+      "loss": 0.104,
+      "step": 5112
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4592180408603038e-05,
+      "loss": 0.1047,
+      "step": 5113
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4587208247251992e-05,
+      "loss": 0.0773,
+      "step": 5114
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.458223613129134e-05,
+      "loss": 0.0993,
+      "step": 5115
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4577264061267824e-05,
+      "loss": 0.1092,
+      "step": 5116
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4572292037728161e-05,
+      "loss": 0.1188,
+      "step": 5117
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4567320061219074e-05,
+      "loss": 0.0981,
+      "step": 5118
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4562348132287284e-05,
+      "loss": 0.1039,
+      "step": 5119
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4557376251479501e-05,
+      "loss": 0.1378,
+      "step": 5120
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4552404419342426e-05,
+      "loss": 0.0882,
+      "step": 5121
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4547432636422764e-05,
+      "loss": 0.1332,
+      "step": 5122
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4542460903267204e-05,
+      "loss": 0.1203,
+      "step": 5123
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4537489220422446e-05,
+      "loss": 0.1376,
+      "step": 5124
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.453251758843517e-05,
+      "loss": 0.1177,
+      "step": 5125
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4527546007852046e-05,
+      "loss": 0.0908,
+      "step": 5126
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.452257447921976e-05,
+      "loss": 0.1306,
+      "step": 5127
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4517603003084966e-05,
+      "loss": 0.1123,
+      "step": 5128
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.451263157999434e-05,
+      "loss": 0.1384,
+      "step": 5129
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4507660210494525e-05,
+      "loss": 0.1312,
+      "step": 5130
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4502688895132174e-05,
+      "loss": 0.1245,
+      "step": 5131
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4497717634453932e-05,
+      "loss": 0.0948,
+      "step": 5132
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4492746429006442e-05,
+      "loss": 0.0941,
+      "step": 5133
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4487775279336323e-05,
+      "loss": 0.1105,
+      "step": 5134
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4482804185990206e-05,
+      "loss": 0.104,
+      "step": 5135
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4477833149514716e-05,
+      "loss": 0.1551,
+      "step": 5136
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4472862170456464e-05,
+      "loss": 0.1227,
+      "step": 5137
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4467891249362047e-05,
+      "loss": 0.0908,
+      "step": 5138
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4462920386778076e-05,
+      "loss": 0.1222,
+      "step": 5139
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.445794958325114e-05,
+      "loss": 0.1136,
+      "step": 5140
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4452978839327831e-05,
+      "loss": 0.0873,
+      "step": 5141
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4448008155554724e-05,
+      "loss": 0.0886,
+      "step": 5142
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4443037532478394e-05,
+      "loss": 0.0812,
+      "step": 5143
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4438066970645412e-05,
+      "loss": 0.1136,
+      "step": 5144
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4433096470602339e-05,
+      "loss": 0.1065,
+      "step": 5145
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4428126032895723e-05,
+      "loss": 0.1509,
+      "step": 5146
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4423155658072116e-05,
+      "loss": 0.1252,
+      "step": 5147
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4418185346678055e-05,
+      "loss": 0.0861,
+      "step": 5148
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4413215099260082e-05,
+      "loss": 0.0977,
+      "step": 5149
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4408244916364712e-05,
+      "loss": 0.091,
+      "step": 5150
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4403274798538467e-05,
+      "loss": 0.0997,
+      "step": 5151
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4398304746327864e-05,
+      "loss": 0.1049,
+      "step": 5152
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4393334760279399e-05,
+      "loss": 0.1186,
+      "step": 5153
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4388364840939585e-05,
+      "loss": 0.0775,
+      "step": 5154
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4383394988854896e-05,
+      "loss": 0.1168,
+      "step": 5155
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4378425204571817e-05,
+      "loss": 0.1433,
+      "step": 5156
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4373455488636832e-05,
+      "loss": 0.202,
+      "step": 5157
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4368485841596406e-05,
+      "loss": 0.1447,
+      "step": 5158
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.436351626399699e-05,
+      "loss": 0.099,
+      "step": 5159
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4358546756385046e-05,
+      "loss": 0.1567,
+      "step": 5160
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4353577319307016e-05,
+      "loss": 0.084,
+      "step": 5161
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4348607953309344e-05,
+      "loss": 0.1057,
+      "step": 5162
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4343638658938444e-05,
+      "loss": 0.0966,
+      "step": 5163
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.433866943674075e-05,
+      "loss": 0.0866,
+      "step": 5164
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4333700287262669e-05,
+      "loss": 0.1167,
+      "step": 5165
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4328731211050616e-05,
+      "loss": 0.0902,
+      "step": 5166
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4323762208650977e-05,
+      "loss": 0.1065,
+      "step": 5167
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4318793280610143e-05,
+      "loss": 0.1596,
+      "step": 5168
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4313824427474502e-05,
+      "loss": 0.1575,
+      "step": 5169
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4308855649790424e-05,
+      "loss": 0.1164,
+      "step": 5170
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4303886948104266e-05,
+      "loss": 0.1212,
+      "step": 5171
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4298918322962398e-05,
+      "loss": 0.152,
+      "step": 5172
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4293949774911154e-05,
+      "loss": 0.0988,
+      "step": 5173
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4288981304496887e-05,
+      "loss": 0.0935,
+      "step": 5174
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4284012912265916e-05,
+      "loss": 0.0893,
+      "step": 5175
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4279044598764567e-05,
+      "loss": 0.0895,
+      "step": 5176
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4274076364539156e-05,
+      "loss": 0.1398,
+      "step": 5177
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.426910821013599e-05,
+      "loss": 0.0846,
+      "step": 5178
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4264140136101353e-05,
+      "loss": 0.0683,
+      "step": 5179
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4259172142981544e-05,
+      "loss": 0.0608,
+      "step": 5180
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4254204231322837e-05,
+      "loss": 0.1462,
+      "step": 5181
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4249236401671505e-05,
+      "loss": 0.1309,
+      "step": 5182
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4244268654573808e-05,
+      "loss": 0.1237,
+      "step": 5183
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4239300990575988e-05,
+      "loss": 0.0923,
+      "step": 5184
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4234333410224294e-05,
+      "loss": 0.0922,
+      "step": 5185
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4229365914064962e-05,
+      "loss": 0.1412,
+      "step": 5186
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4224398502644216e-05,
+      "loss": 0.1664,
+      "step": 5187
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.421943117650826e-05,
+      "loss": 0.0809,
+      "step": 5188
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.421446393620331e-05,
+      "loss": 0.1013,
+      "step": 5189
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4209496782275555e-05,
+      "loss": 0.1003,
+      "step": 5190
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4204529715271187e-05,
+      "loss": 0.1143,
+      "step": 5191
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4199562735736377e-05,
+      "loss": 0.1818,
+      "step": 5192
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4194595844217292e-05,
+      "loss": 0.0871,
+      "step": 5193
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4189629041260093e-05,
+      "loss": 0.0567,
+      "step": 5194
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4184662327410929e-05,
+      "loss": 0.1391,
+      "step": 5195
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4179695703215927e-05,
+      "loss": 0.1539,
+      "step": 5196
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4174729169221224e-05,
+      "loss": 0.0623,
+      "step": 5197
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4169762725972936e-05,
+      "loss": 0.0886,
+      "step": 5198
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4164796374017174e-05,
+      "loss": 0.1092,
+      "step": 5199
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.415983011390003e-05,
+      "loss": 0.049,
+      "step": 5200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4154863946167594e-05,
+      "loss": 0.151,
+      "step": 5201
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4149897871365942e-05,
+      "loss": 0.1445,
+      "step": 5202
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.414493189004115e-05,
+      "loss": 0.0958,
+      "step": 5203
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4139966002739263e-05,
+      "loss": 0.0826,
+      "step": 5204
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4135000210006333e-05,
+      "loss": 0.1021,
+      "step": 5205
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4130034512388395e-05,
+      "loss": 0.0608,
+      "step": 5206
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4125068910431483e-05,
+      "loss": 0.1561,
+      "step": 5207
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4120103404681602e-05,
+      "loss": 0.1251,
+      "step": 5208
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4115137995684757e-05,
+      "loss": 0.124,
+      "step": 5209
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.411017268398695e-05,
+      "loss": 0.088,
+      "step": 5210
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4105207470134163e-05,
+      "loss": 0.0617,
+      "step": 5211
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.410024235467236e-05,
+      "loss": 0.1353,
+      "step": 5212
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4095277338147508e-05,
+      "loss": 0.1223,
+      "step": 5213
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4090312421105563e-05,
+      "loss": 0.1155,
+      "step": 5214
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4085347604092457e-05,
+      "loss": 0.1061,
+      "step": 5215
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.408038288765413e-05,
+      "loss": 0.1106,
+      "step": 5216
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.407541827233649e-05,
+      "loss": 0.0646,
+      "step": 5217
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4070453758685446e-05,
+      "loss": 0.0545,
+      "step": 5218
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4065489347246894e-05,
+      "loss": 0.1619,
+      "step": 5219
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4060525038566723e-05,
+      "loss": 0.0892,
+      "step": 5220
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4055560833190798e-05,
+      "loss": 0.1339,
+      "step": 5221
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4050596731664988e-05,
+      "loss": 0.0974,
+      "step": 5222
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4045632734535136e-05,
+      "loss": 0.0959,
+      "step": 5223
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4040668842347093e-05,
+      "loss": 0.0783,
+      "step": 5224
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4035705055646675e-05,
+      "loss": 0.1135,
+      "step": 5225
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.40307413749797e-05,
+      "loss": 0.1145,
+      "step": 5226
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4025777800891975e-05,
+      "loss": 0.1387,
+      "step": 5227
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4020814333929295e-05,
+      "loss": 0.1134,
+      "step": 5228
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.401585097463743e-05,
+      "loss": 0.088,
+      "step": 5229
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4010887723562158e-05,
+      "loss": 0.1285,
+      "step": 5230
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4005924581249235e-05,
+      "loss": 0.0756,
+      "step": 5231
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4000961548244406e-05,
+      "loss": 0.113,
+      "step": 5232
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.39959986250934e-05,
+      "loss": 0.0425,
+      "step": 5233
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3991035812341938e-05,
+      "loss": 0.1445,
+      "step": 5234
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3986073110535734e-05,
+      "loss": 0.1161,
+      "step": 5235
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3981110520220485e-05,
+      "loss": 0.0469,
+      "step": 5236
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3976148041941866e-05,
+      "loss": 0.1733,
+      "step": 5237
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3971185676245552e-05,
+      "loss": 0.0876,
+      "step": 5238
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3966223423677208e-05,
+      "loss": 0.0601,
+      "step": 5239
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3961261284782485e-05,
+      "loss": 0.1918,
+      "step": 5240
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3956299260107e-05,
+      "loss": 0.1179,
+      "step": 5241
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3951337350196391e-05,
+      "loss": 0.0808,
+      "step": 5242
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.394637555559626e-05,
+      "loss": 0.0874,
+      "step": 5243
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3941413876852209e-05,
+      "loss": 0.1693,
+      "step": 5244
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3936452314509821e-05,
+      "loss": 0.0825,
+      "step": 5245
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3931490869114659e-05,
+      "loss": 0.1357,
+      "step": 5246
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.392652954121229e-05,
+      "loss": 0.1024,
+      "step": 5247
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3921568331348254e-05,
+      "loss": 0.1126,
+      "step": 5248
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3916607240068093e-05,
+      "loss": 0.1423,
+      "step": 5249
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3911646267917317e-05,
+      "loss": 0.1198,
+      "step": 5250
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3906685415441432e-05,
+      "loss": 0.1139,
+      "step": 5251
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3901724683185935e-05,
+      "loss": 0.1073,
+      "step": 5252
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.389676407169631e-05,
+      "loss": 0.0419,
+      "step": 5253
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3891803581518011e-05,
+      "loss": 0.1129,
+      "step": 5254
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3886843213196504e-05,
+      "loss": 0.1039,
+      "step": 5255
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.388188296727722e-05,
+      "loss": 0.0867,
+      "step": 5256
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3876922844305595e-05,
+      "loss": 0.1428,
+      "step": 5257
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3871962844827034e-05,
+      "loss": 0.1077,
+      "step": 5258
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3867002969386936e-05,
+      "loss": 0.1384,
+      "step": 5259
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3862043218530692e-05,
+      "loss": 0.1448,
+      "step": 5260
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3857083592803674e-05,
+      "loss": 0.0986,
+      "step": 5261
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3852124092751232e-05,
+      "loss": 0.0659,
+      "step": 5262
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3847164718918716e-05,
+      "loss": 0.1361,
+      "step": 5263
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3842205471851458e-05,
+      "loss": 0.0981,
+      "step": 5264
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3837246352094777e-05,
+      "loss": 0.1191,
+      "step": 5265
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3832287360193967e-05,
+      "loss": 0.1391,
+      "step": 5266
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3827328496694322e-05,
+      "loss": 0.1224,
+      "step": 5267
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3822369762141115e-05,
+      "loss": 0.0723,
+      "step": 5268
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3817411157079612e-05,
+      "loss": 0.158,
+      "step": 5269
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3812452682055052e-05,
+      "loss": 0.0842,
+      "step": 5270
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3807494337612664e-05,
+      "loss": 0.0798,
+      "step": 5271
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3802536124297674e-05,
+      "loss": 0.1047,
+      "step": 5272
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3797578042655278e-05,
+      "loss": 0.0957,
+      "step": 5273
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3792620093230674e-05,
+      "loss": 0.1073,
+      "step": 5274
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3787662276569027e-05,
+      "loss": 0.1198,
+      "step": 5275
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3782704593215491e-05,
+      "loss": 0.1435,
+      "step": 5276
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3777747043715227e-05,
+      "loss": 0.1678,
+      "step": 5277
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3772789628613358e-05,
+      "loss": 0.0995,
+      "step": 5278
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.376783234845499e-05,
+      "loss": 0.0739,
+      "step": 5279
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3762875203785234e-05,
+      "loss": 0.0935,
+      "step": 5280
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3757918195149171e-05,
+      "loss": 0.1066,
+      "step": 5281
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3752961323091878e-05,
+      "loss": 0.0704,
+      "step": 5282
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3748004588158402e-05,
+      "loss": 0.0996,
+      "step": 5283
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3743047990893783e-05,
+      "loss": 0.0639,
+      "step": 5284
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3738091531843053e-05,
+      "loss": 0.1139,
+      "step": 5285
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3733135211551222e-05,
+      "loss": 0.0911,
+      "step": 5286
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3728179030563276e-05,
+      "loss": 0.1709,
+      "step": 5287
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3723222989424203e-05,
+      "loss": 0.0798,
+      "step": 5288
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3718267088678961e-05,
+      "loss": 0.1251,
+      "step": 5289
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3713311328872508e-05,
+      "loss": 0.1163,
+      "step": 5290
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3708355710549765e-05,
+      "loss": 0.1677,
+      "step": 5291
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3703400234255657e-05,
+      "loss": 0.1682,
+      "step": 5292
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3698444900535083e-05,
+      "loss": 0.135,
+      "step": 5293
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3693489709932938e-05,
+      "loss": 0.0798,
+      "step": 5294
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3688534662994081e-05,
+      "loss": 0.1048,
+      "step": 5295
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3683579760263368e-05,
+      "loss": 0.0793,
+      "step": 5296
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3678625002285645e-05,
+      "loss": 0.0888,
+      "step": 5297
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3673670389605735e-05,
+      "loss": 0.098,
+      "step": 5298
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3668715922768433e-05,
+      "loss": 0.1582,
+      "step": 5299
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3663761602318542e-05,
+      "loss": 0.0817,
+      "step": 5300
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3658807428800829e-05,
+      "loss": 0.0718,
+      "step": 5301
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3653853402760063e-05,
+      "loss": 0.1467,
+      "step": 5302
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3648899524740978e-05,
+      "loss": 0.1627,
+      "step": 5303
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3643945795288299e-05,
+      "loss": 0.1259,
+      "step": 5304
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3638992214946742e-05,
+      "loss": 0.119,
+      "step": 5305
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3634038784260995e-05,
+      "loss": 0.1422,
+      "step": 5306
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3629085503775744e-05,
+      "loss": 0.0968,
+      "step": 5307
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3624132374035641e-05,
+      "loss": 0.1158,
+      "step": 5308
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.361917939558533e-05,
+      "loss": 0.1323,
+      "step": 5309
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3614226568969441e-05,
+      "loss": 0.0993,
+      "step": 5310
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.360927389473259e-05,
+      "loss": 0.1392,
+      "step": 5311
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3604321373419357e-05,
+      "loss": 0.1131,
+      "step": 5312
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3599369005574331e-05,
+      "loss": 0.1269,
+      "step": 5313
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3594416791742065e-05,
+      "loss": 0.1111,
+      "step": 5314
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3589464732467112e-05,
+      "loss": 0.0657,
+      "step": 5315
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3584512828293984e-05,
+      "loss": 0.1102,
+      "step": 5316
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3579561079767205e-05,
+      "loss": 0.1008,
+      "step": 5317
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3574609487431254e-05,
+      "loss": 0.1207,
+      "step": 5318
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.356965805183062e-05,
+      "loss": 0.1053,
+      "step": 5319
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3564706773509748e-05,
+      "loss": 0.1216,
+      "step": 5320
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.355975565301308e-05,
+      "loss": 0.0672,
+      "step": 5321
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3554804690885046e-05,
+      "loss": 0.0759,
+      "step": 5322
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.354985388767005e-05,
+      "loss": 0.1251,
+      "step": 5323
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3544903243912473e-05,
+      "loss": 0.0843,
+      "step": 5324
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.353995276015669e-05,
+      "loss": 0.1232,
+      "step": 5325
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3535002436947056e-05,
+      "loss": 0.1106,
+      "step": 5326
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3530052274827908e-05,
+      "loss": 0.0912,
+      "step": 5327
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3525102274343557e-05,
+      "loss": 0.091,
+      "step": 5328
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3520152436038304e-05,
+      "loss": 0.1185,
+      "step": 5329
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3515202760456437e-05,
+      "loss": 0.1197,
+      "step": 5330
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3510253248142218e-05,
+      "loss": 0.0948,
+      "step": 5331
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3505303899639886e-05,
+      "loss": 0.1316,
+      "step": 5332
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3500354715493678e-05,
+      "loss": 0.0976,
+      "step": 5333
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3495405696247798e-05,
+      "loss": 0.1198,
+      "step": 5334
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3490456842446442e-05,
+      "loss": 0.1166,
+      "step": 5335
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.348550815463379e-05,
+      "loss": 0.179,
+      "step": 5336
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3480559633353979e-05,
+      "loss": 0.1128,
+      "step": 5337
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3475611279151162e-05,
+      "loss": 0.0645,
+      "step": 5338
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3470663092569453e-05,
+      "loss": 0.1533,
+      "step": 5339
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3465715074152954e-05,
+      "loss": 0.1483,
+      "step": 5340
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3460767224445741e-05,
+      "loss": 0.1168,
+      "step": 5341
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3455819543991885e-05,
+      "loss": 0.0971,
+      "step": 5342
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3450872033335424e-05,
+      "loss": 0.1028,
+      "step": 5343
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3445924693020396e-05,
+      "loss": 0.0847,
+      "step": 5344
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3440977523590795e-05,
+      "loss": 0.1158,
+      "step": 5345
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3436030525590612e-05,
+      "loss": 0.1623,
+      "step": 5346
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3431083699563821e-05,
+      "loss": 0.0871,
+      "step": 5347
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3426137046054375e-05,
+      "loss": 0.0639,
+      "step": 5348
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3421190565606197e-05,
+      "loss": 0.0916,
+      "step": 5349
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3416244258763206e-05,
+      "loss": 0.1528,
+      "step": 5350
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.341129812606929e-05,
+      "loss": 0.0951,
+      "step": 5351
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3406352168068336e-05,
+      "loss": 0.126,
+      "step": 5352
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3401406385304188e-05,
+      "loss": 0.1371,
+      "step": 5353
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3396460778320683e-05,
+      "loss": 0.1222,
+      "step": 5354
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3391515347661642e-05,
+      "loss": 0.1011,
+      "step": 5355
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3386570093870864e-05,
+      "loss": 0.1043,
+      "step": 5356
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3381625017492117e-05,
+      "loss": 0.073,
+      "step": 5357
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3376680119069168e-05,
+      "loss": 0.08,
+      "step": 5358
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3371735399145752e-05,
+      "loss": 0.0801,
+      "step": 5359
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3366790858265597e-05,
+      "loss": 0.1415,
+      "step": 5360
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.336184649697239e-05,
+      "loss": 0.1043,
+      "step": 5361
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3356902315809813e-05,
+      "loss": 0.0912,
+      "step": 5362
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3351958315321534e-05,
+      "loss": 0.1269,
+      "step": 5363
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3347014496051182e-05,
+      "loss": 0.0894,
+      "step": 5364
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3342070858542394e-05,
+      "loss": 0.0812,
+      "step": 5365
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.333712740333875e-05,
+      "loss": 0.1018,
+      "step": 5366
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3332184130983842e-05,
+      "loss": 0.1672,
+      "step": 5367
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3327241042021227e-05,
+      "loss": 0.1022,
+      "step": 5368
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3322298136994453e-05,
+      "loss": 0.1107,
+      "step": 5369
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3317355416447026e-05,
+      "loss": 0.056,
+      "step": 5370
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.331241288092245e-05,
+      "loss": 0.1049,
+      "step": 5371
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.330747053096421e-05,
+      "loss": 0.1049,
+      "step": 5372
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.330252836711576e-05,
+      "loss": 0.1115,
+      "step": 5373
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3297586389920536e-05,
+      "loss": 0.1003,
+      "step": 5374
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3292644599921958e-05,
+      "loss": 0.0991,
+      "step": 5375
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3287702997663422e-05,
+      "loss": 0.0922,
+      "step": 5376
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.328276158368831e-05,
+      "loss": 0.1155,
+      "step": 5377
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3277820358539971e-05,
+      "loss": 0.1289,
+      "step": 5378
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3272879322761739e-05,
+      "loss": 0.0759,
+      "step": 5379
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3267938476896935e-05,
+      "loss": 0.1401,
+      "step": 5380
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3262997821488851e-05,
+      "loss": 0.0969,
+      "step": 5381
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3258057357080749e-05,
+      "loss": 0.109,
+      "step": 5382
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3253117084215892e-05,
+      "loss": 0.136,
+      "step": 5383
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3248177003437504e-05,
+      "loss": 0.0917,
+      "step": 5384
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.32432371152888e-05,
+      "loss": 0.0594,
+      "step": 5385
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3238297420312965e-05,
+      "loss": 0.0844,
+      "step": 5386
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3233357919053157e-05,
+      "loss": 0.1404,
+      "step": 5387
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3228418612052536e-05,
+      "loss": 0.1279,
+      "step": 5388
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3223479499854223e-05,
+      "loss": 0.1609,
+      "step": 5389
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3218540583001305e-05,
+      "loss": 0.0803,
+      "step": 5390
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3213601862036883e-05,
+      "loss": 0.1103,
+      "step": 5391
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3208663337504005e-05,
+      "loss": 0.0807,
+      "step": 5392
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3203725009945717e-05,
+      "loss": 0.1429,
+      "step": 5393
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3198786879905025e-05,
+      "loss": 0.1138,
+      "step": 5394
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3193848947924932e-05,
+      "loss": 0.0867,
+      "step": 5395
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3188911214548406e-05,
+      "loss": 0.1041,
+      "step": 5396
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3183973680318404e-05,
+      "loss": 0.1342,
+      "step": 5397
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3179036345777854e-05,
+      "loss": 0.1389,
+      "step": 5398
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3174099211469653e-05,
+      "loss": 0.1268,
+      "step": 5399
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3169162277936698e-05,
+      "loss": 0.1279,
+      "step": 5400
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3164225545721842e-05,
+      "loss": 0.0985,
+      "step": 5401
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3159289015367938e-05,
+      "loss": 0.0948,
+      "step": 5402
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3154352687417795e-05,
+      "loss": 0.1493,
+      "step": 5403
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.314941656241421e-05,
+      "loss": 0.0567,
+      "step": 5404
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3144480640899964e-05,
+      "loss": 0.1211,
+      "step": 5405
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3139544923417806e-05,
+      "loss": 0.0994,
+      "step": 5406
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3134609410510458e-05,
+      "loss": 0.1036,
+      "step": 5407
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3129674102720635e-05,
+      "loss": 0.1641,
+      "step": 5408
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3124739000591015e-05,
+      "loss": 0.102,
+      "step": 5409
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3119804104664272e-05,
+      "loss": 0.1168,
+      "step": 5410
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.311486941548303e-05,
+      "loss": 0.1293,
+      "step": 5411
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.310993493358991e-05,
+      "loss": 0.1116,
+      "step": 5412
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.310500065952751e-05,
+      "loss": 0.1534,
+      "step": 5413
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3100066593838401e-05,
+      "loss": 0.1381,
+      "step": 5414
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3095132737065121e-05,
+      "loss": 0.1096,
+      "step": 5415
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3090199089750203e-05,
+      "loss": 0.073,
+      "step": 5416
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3085265652436146e-05,
+      "loss": 0.1043,
+      "step": 5417
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3080332425665433e-05,
+      "loss": 0.0927,
+      "step": 5418
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.307539940998051e-05,
+      "loss": 0.0958,
+      "step": 5419
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.307046660592382e-05,
+      "loss": 0.1134,
+      "step": 5420
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.306553401403776e-05,
+      "loss": 0.1342,
+      "step": 5421
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.306060163486473e-05,
+      "loss": 0.0925,
+      "step": 5422
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3055669468947082e-05,
+      "loss": 0.1255,
+      "step": 5423
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3050737516827154e-05,
+      "loss": 0.0935,
+      "step": 5424
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3045805779047269e-05,
+      "loss": 0.0845,
+      "step": 5425
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3040874256149709e-05,
+      "loss": 0.1264,
+      "step": 5426
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3035942948676755e-05,
+      "loss": 0.1308,
+      "step": 5427
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.303101185717064e-05,
+      "loss": 0.0771,
+      "step": 5428
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3026080982173587e-05,
+      "loss": 0.136,
+      "step": 5429
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3021150324227797e-05,
+      "loss": 0.0823,
+      "step": 5430
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3016219883875443e-05,
+      "loss": 0.118,
+      "step": 5431
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3011289661658665e-05,
+      "loss": 0.124,
+      "step": 5432
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3006359658119596e-05,
+      "loss": 0.0964,
+      "step": 5433
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3001429873800335e-05,
+      "loss": 0.1415,
+      "step": 5434
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2996500309242964e-05,
+      "loss": 0.1105,
+      "step": 5435
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2991570964989528e-05,
+      "loss": 0.0785,
+      "step": 5436
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2986641841582056e-05,
+      "loss": 0.1654,
+      "step": 5437
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2981712939562558e-05,
+      "loss": 0.1097,
+      "step": 5438
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2976784259473013e-05,
+      "loss": 0.0978,
+      "step": 5439
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2971855801855368e-05,
+      "loss": 0.1844,
+      "step": 5440
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2966927567251561e-05,
+      "loss": 0.1134,
+      "step": 5441
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2961999556203497e-05,
+      "loss": 0.0713,
+      "step": 5442
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2957071769253067e-05,
+      "loss": 0.0636,
+      "step": 5443
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2952144206942109e-05,
+      "loss": 0.1009,
+      "step": 5444
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2947216869812471e-05,
+      "loss": 0.1689,
+      "step": 5445
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2942289758405953e-05,
+      "loss": 0.1383,
+      "step": 5446
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2937362873264348e-05,
+      "loss": 0.0709,
+      "step": 5447
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2932436214929405e-05,
+      "loss": 0.2161,
+      "step": 5448
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2927509783942854e-05,
+      "loss": 0.0679,
+      "step": 5449
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2922583580846414e-05,
+      "loss": 0.083,
+      "step": 5450
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2917657606181765e-05,
+      "loss": 0.1179,
+      "step": 5451
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2912731860490557e-05,
+      "loss": 0.0753,
+      "step": 5452
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.290780634431443e-05,
+      "loss": 0.1298,
+      "step": 5453
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2902881058194986e-05,
+      "loss": 0.1173,
+      "step": 5454
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2897956002673816e-05,
+      "loss": 0.0553,
+      "step": 5455
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2893031178292477e-05,
+      "loss": 0.1445,
+      "step": 5456
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2888106585592488e-05,
+      "loss": 0.0854,
+      "step": 5457
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2883182225115365e-05,
+      "loss": 0.1324,
+      "step": 5458
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2878258097402584e-05,
+      "loss": 0.0951,
+      "step": 5459
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2873334202995609e-05,
+      "loss": 0.14,
+      "step": 5460
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2868410542435858e-05,
+      "loss": 0.1008,
+      "step": 5461
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.286348711626474e-05,
+      "loss": 0.1076,
+      "step": 5462
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2858563925023631e-05,
+      "loss": 0.1186,
+      "step": 5463
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2853640969253887e-05,
+      "loss": 0.158,
+      "step": 5464
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2848718249496825e-05,
+      "loss": 0.1379,
+      "step": 5465
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2843795766293755e-05,
+      "loss": 0.1043,
+      "step": 5466
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2838873520185941e-05,
+      "loss": 0.1057,
+      "step": 5467
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2833951511714645e-05,
+      "loss": 0.0684,
+      "step": 5468
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2829029741421078e-05,
+      "loss": 0.12,
+      "step": 5469
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2824108209846434e-05,
+      "loss": 0.1186,
+      "step": 5470
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2819186917531887e-05,
+      "loss": 0.0841,
+      "step": 5471
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2814265865018587e-05,
+      "loss": 0.1197,
+      "step": 5472
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.280934505284764e-05,
+      "loss": 0.1595,
+      "step": 5473
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2804424481560138e-05,
+      "loss": 0.0753,
+      "step": 5474
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2799504151697149e-05,
+      "loss": 0.1065,
+      "step": 5475
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2794584063799712e-05,
+      "loss": 0.0772,
+      "step": 5476
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2789664218408825e-05,
+      "loss": 0.1794,
+      "step": 5477
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2784744616065486e-05,
+      "loss": 0.056,
+      "step": 5478
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2779825257310645e-05,
+      "loss": 0.046,
+      "step": 5479
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2774906142685241e-05,
+      "loss": 0.0983,
+      "step": 5480
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2769987272730166e-05,
+      "loss": 0.1349,
+      "step": 5481
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.27650686479863e-05,
+      "loss": 0.0758,
+      "step": 5482
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2760150268994498e-05,
+      "loss": 0.0808,
+      "step": 5483
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2755232136295577e-05,
+      "loss": 0.1159,
+      "step": 5484
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2750314250430341e-05,
+      "loss": 0.0872,
+      "step": 5485
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2745396611939551e-05,
+      "loss": 0.1124,
+      "step": 5486
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2740479221363946e-05,
+      "loss": 0.0837,
+      "step": 5487
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2735562079244247e-05,
+      "loss": 0.1358,
+      "step": 5488
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2730645186121143e-05,
+      "loss": 0.0995,
+      "step": 5489
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2725728542535281e-05,
+      "loss": 0.0682,
+      "step": 5490
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2720812149027305e-05,
+      "loss": 0.089,
+      "step": 5491
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.271589600613781e-05,
+      "loss": 0.0513,
+      "step": 5492
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2710980114407383e-05,
+      "loss": 0.0996,
+      "step": 5493
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2706064474376567e-05,
+      "loss": 0.1124,
+      "step": 5494
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.270114908658588e-05,
+      "loss": 0.1734,
+      "step": 5495
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2696233951575823e-05,
+      "loss": 0.0982,
+      "step": 5496
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2691319069886864e-05,
+      "loss": 0.1336,
+      "step": 5497
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2686404442059435e-05,
+      "loss": 0.1432,
+      "step": 5498
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2681490068633943e-05,
+      "loss": 0.1342,
+      "step": 5499
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.267657595015078e-05,
+      "loss": 0.0902,
+      "step": 5500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2671662087150299e-05,
+      "loss": 0.0587,
+      "step": 5501
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2666748480172817e-05,
+      "loss": 0.0605,
+      "step": 5502
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.266183512975864e-05,
+      "loss": 0.0827,
+      "step": 5503
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2656922036448032e-05,
+      "loss": 0.093,
+      "step": 5504
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2652009200781245e-05,
+      "loss": 0.0844,
+      "step": 5505
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2647096623298485e-05,
+      "loss": 0.1086,
+      "step": 5506
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2642184304539932e-05,
+      "loss": 0.1371,
+      "step": 5507
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2637272245045751e-05,
+      "loss": 0.1307,
+      "step": 5508
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.263236044535607e-05,
+      "loss": 0.0873,
+      "step": 5509
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2627448906010979e-05,
+      "loss": 0.1514,
+      "step": 5510
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2622537627550557e-05,
+      "loss": 0.0985,
+      "step": 5511
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2617626610514842e-05,
+      "loss": 0.0919,
+      "step": 5512
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2612715855443855e-05,
+      "loss": 0.1895,
+      "step": 5513
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2607805362877571e-05,
+      "loss": 0.066,
+      "step": 5514
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2602895133355946e-05,
+      "loss": 0.0993,
+      "step": 5515
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2597985167418912e-05,
+      "loss": 0.0876,
+      "step": 5516
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2593075465606362e-05,
+      "loss": 0.1152,
+      "step": 5517
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2588166028458174e-05,
+      "loss": 0.1022,
+      "step": 5518
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2583256856514177e-05,
+      "loss": 0.0596,
+      "step": 5519
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.257834795031418e-05,
+      "loss": 0.0793,
+      "step": 5520
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.257343931039797e-05,
+      "loss": 0.1006,
+      "step": 5521
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2568530937305306e-05,
+      "loss": 0.0577,
+      "step": 5522
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2563622831575898e-05,
+      "loss": 0.1151,
+      "step": 5523
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2558714993749439e-05,
+      "loss": 0.1571,
+      "step": 5524
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2553807424365603e-05,
+      "loss": 0.1592,
+      "step": 5525
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2548900123964018e-05,
+      "loss": 0.0914,
+      "step": 5526
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2543993093084287e-05,
+      "loss": 0.1196,
+      "step": 5527
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2539086332265987e-05,
+      "loss": 0.1202,
+      "step": 5528
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2534179842048662e-05,
+      "loss": 0.1093,
+      "step": 5529
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2529273622971833e-05,
+      "loss": 0.0982,
+      "step": 5530
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2524367675574979e-05,
+      "loss": 0.0721,
+      "step": 5531
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2519462000397555e-05,
+      "loss": 0.0785,
+      "step": 5532
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2514556597978994e-05,
+      "loss": 0.1212,
+      "step": 5533
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.250965146885869e-05,
+      "loss": 0.1469,
+      "step": 5534
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2504746613576003e-05,
+      "loss": 0.1584,
+      "step": 5535
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2499842032670273e-05,
+      "loss": 0.1735,
+      "step": 5536
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2494937726680801e-05,
+      "loss": 0.0951,
+      "step": 5537
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2490033696146873e-05,
+      "loss": 0.1086,
+      "step": 5538
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2485129941607724e-05,
+      "loss": 0.1374,
+      "step": 5539
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2480226463602573e-05,
+      "loss": 0.0444,
+      "step": 5540
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.24753232626706e-05,
+      "loss": 0.0973,
+      "step": 5541
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.247042033935097e-05,
+      "loss": 0.1519,
+      "step": 5542
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.246551769418279e-05,
+      "loss": 0.0708,
+      "step": 5543
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2460615327705165e-05,
+      "loss": 0.091,
+      "step": 5544
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.245571324045715e-05,
+      "loss": 0.1996,
+      "step": 5545
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2450811432977777e-05,
+      "loss": 0.0988,
+      "step": 5546
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2445909905806056e-05,
+      "loss": 0.1466,
+      "step": 5547
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2441008659480949e-05,
+      "loss": 0.1103,
+      "step": 5548
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.243610769454139e-05,
+      "loss": 0.1087,
+      "step": 5549
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2431207011526297e-05,
+      "loss": 0.1166,
+      "step": 5550
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2426306610974546e-05,
+      "loss": 0.1153,
+      "step": 5551
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2421406493424972e-05,
+      "loss": 0.0847,
+      "step": 5552
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.24165066594164e-05,
+      "loss": 0.0962,
+      "step": 5553
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2411607109487608e-05,
+      "loss": 0.1123,
+      "step": 5554
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2406707844177359e-05,
+      "loss": 0.1728,
+      "step": 5555
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.240180886402436e-05,
+      "loss": 0.1155,
+      "step": 5556
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2396910169567304e-05,
+      "loss": 0.0973,
+      "step": 5557
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2392011761344857e-05,
+      "loss": 0.108,
+      "step": 5558
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2387113639895644e-05,
+      "loss": 0.1478,
+      "step": 5559
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2382215805758249e-05,
+      "loss": 0.1208,
+      "step": 5560
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2377318259471247e-05,
+      "loss": 0.1579,
+      "step": 5561
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2372421001573165e-05,
+      "loss": 0.1346,
+      "step": 5562
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.236752403260251e-05,
+      "loss": 0.1016,
+      "step": 5563
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2362627353097744e-05,
+      "loss": 0.1331,
+      "step": 5564
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.23577309635973e-05,
+      "loss": 0.1389,
+      "step": 5565
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2352834864639592e-05,
+      "loss": 0.154,
+      "step": 5566
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2347939056762994e-05,
+      "loss": 0.1007,
+      "step": 5567
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2343043540505835e-05,
+      "loss": 0.0657,
+      "step": 5568
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2338148316406431e-05,
+      "loss": 0.0736,
+      "step": 5569
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2333253385003057e-05,
+      "loss": 0.1203,
+      "step": 5570
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2328358746833966e-05,
+      "loss": 0.1204,
+      "step": 5571
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2323464402437359e-05,
+      "loss": 0.1333,
+      "step": 5572
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2318570352351417e-05,
+      "loss": 0.1143,
+      "step": 5573
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.231367659711429e-05,
+      "loss": 0.1424,
+      "step": 5574
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2308783137264098e-05,
+      "loss": 0.1471,
+      "step": 5575
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2303889973338922e-05,
+      "loss": 0.139,
+      "step": 5576
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2298997105876803e-05,
+      "loss": 0.134,
+      "step": 5577
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.229410453541577e-05,
+      "loss": 0.0825,
+      "step": 5578
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2289212262493798e-05,
+      "loss": 0.1028,
+      "step": 5579
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2284320287648852e-05,
+      "loss": 0.1152,
+      "step": 5580
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2279428611418837e-05,
+      "loss": 0.0759,
+      "step": 5581
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2274537234341645e-05,
+      "loss": 0.1334,
+      "step": 5582
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2269646156955135e-05,
+      "loss": 0.1482,
+      "step": 5583
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2264755379797124e-05,
+      "loss": 0.0812,
+      "step": 5584
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2259864903405395e-05,
+      "loss": 0.0504,
+      "step": 5585
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.225497472831771e-05,
+      "loss": 0.0821,
+      "step": 5586
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2250084855071782e-05,
+      "loss": 0.0964,
+      "step": 5587
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2245195284205314e-05,
+      "loss": 0.1033,
+      "step": 5588
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2240306016255945e-05,
+      "loss": 0.1074,
+      "step": 5589
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2235417051761302e-05,
+      "loss": 0.1518,
+      "step": 5590
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2230528391258978e-05,
+      "loss": 0.0574,
+      "step": 5591
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2225640035286527e-05,
+      "loss": 0.1418,
+      "step": 5592
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2220751984381462e-05,
+      "loss": 0.0738,
+      "step": 5593
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2215864239081281e-05,
+      "loss": 0.0778,
+      "step": 5594
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2210976799923428e-05,
+      "loss": 0.1179,
+      "step": 5595
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2206089667445338e-05,
+      "loss": 0.0445,
+      "step": 5596
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2201202842184388e-05,
+      "loss": 0.1035,
+      "step": 5597
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.219631632467793e-05,
+      "loss": 0.0893,
+      "step": 5598
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2191430115463285e-05,
+      "loss": 0.1015,
+      "step": 5599
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2186544215077747e-05,
+      "loss": 0.0846,
+      "step": 5600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2181658624058555e-05,
+      "loss": 0.0664,
+      "step": 5601
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2176773342942928e-05,
+      "loss": 0.1313,
+      "step": 5602
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.217188837226806e-05,
+      "loss": 0.0643,
+      "step": 5603
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2167003712571094e-05,
+      "loss": 0.1161,
+      "step": 5604
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2162119364389136e-05,
+      "loss": 0.1115,
+      "step": 5605
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2157235328259279e-05,
+      "loss": 0.1149,
+      "step": 5606
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2152351604718564e-05,
+      "loss": 0.1034,
+      "step": 5607
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2147468194304006e-05,
+      "loss": 0.0531,
+      "step": 5608
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2142585097552585e-05,
+      "loss": 0.1266,
+      "step": 5609
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2137702315001232e-05,
+      "loss": 0.1402,
+      "step": 5610
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2132819847186869e-05,
+      "loss": 0.1307,
+      "step": 5611
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2127937694646362e-05,
+      "loss": 0.0807,
+      "step": 5612
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2123055857916558e-05,
+      "loss": 0.1463,
+      "step": 5613
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2118174337534255e-05,
+      "loss": 0.1644,
+      "step": 5614
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.211329313403622e-05,
+      "loss": 0.0659,
+      "step": 5615
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2108412247959195e-05,
+      "loss": 0.1672,
+      "step": 5616
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2103531679839882e-05,
+      "loss": 0.1185,
+      "step": 5617
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2098651430214936e-05,
+      "loss": 0.1083,
+      "step": 5618
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2093771499620996e-05,
+      "loss": 0.067,
+      "step": 5619
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2088891888594649e-05,
+      "loss": 0.105,
+      "step": 5620
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2084012597672468e-05,
+      "loss": 0.1317,
+      "step": 5621
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2079133627390964e-05,
+      "loss": 0.093,
+      "step": 5622
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2074254978286631e-05,
+      "loss": 0.1428,
+      "step": 5623
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2069376650895922e-05,
+      "loss": 0.1057,
+      "step": 5624
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2064498645755265e-05,
+      "loss": 0.112,
+      "step": 5625
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.205962096340103e-05,
+      "loss": 0.078,
+      "step": 5626
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2054743604369569e-05,
+      "loss": 0.1251,
+      "step": 5627
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2049866569197201e-05,
+      "loss": 0.1326,
+      "step": 5628
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2044989858420198e-05,
+      "loss": 0.1499,
+      "step": 5629
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2040113472574796e-05,
+      "loss": 0.1415,
+      "step": 5630
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2035237412197207e-05,
+      "loss": 0.083,
+      "step": 5631
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2030361677823595e-05,
+      "loss": 0.1003,
+      "step": 5632
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2025486269990102e-05,
+      "loss": 0.0633,
+      "step": 5633
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.202061118923282e-05,
+      "loss": 0.0966,
+      "step": 5634
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2015736436087806e-05,
+      "loss": 0.101,
+      "step": 5635
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2010862011091093e-05,
+      "loss": 0.1151,
+      "step": 5636
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2005987914778667e-05,
+      "loss": 0.1516,
+      "step": 5637
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2001114147686487e-05,
+      "loss": 0.1027,
+      "step": 5638
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1996240710350466e-05,
+      "loss": 0.1139,
+      "step": 5639
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.199136760330648e-05,
+      "loss": 0.1132,
+      "step": 5640
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1986494827090381e-05,
+      "loss": 0.1036,
+      "step": 5641
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1981622382237982e-05,
+      "loss": 0.108,
+      "step": 5642
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.197675026928504e-05,
+      "loss": 0.14,
+      "step": 5643
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.19718784887673e-05,
+      "loss": 0.1049,
+      "step": 5644
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1967007041220459e-05,
+      "loss": 0.1083,
+      "step": 5645
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1962135927180181e-05,
+      "loss": 0.1465,
+      "step": 5646
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.195726514718209e-05,
+      "loss": 0.0911,
+      "step": 5647
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1952394701761772e-05,
+      "loss": 0.1188,
+      "step": 5648
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1947524591454782e-05,
+      "loss": 0.1113,
+      "step": 5649
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1942654816796643e-05,
+      "loss": 0.1193,
+      "step": 5650
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1937785378322817e-05,
+      "loss": 0.1255,
+      "step": 5651
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1932916276568752e-05,
+      "loss": 0.1459,
+      "step": 5652
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1928047512069859e-05,
+      "loss": 0.1441,
+      "step": 5653
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1923179085361503e-05,
+      "loss": 0.0843,
+      "step": 5654
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1918310996979002e-05,
+      "loss": 0.0763,
+      "step": 5655
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1913443247457665e-05,
+      "loss": 0.1362,
+      "step": 5656
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1908575837332734e-05,
+      "loss": 0.0843,
+      "step": 5657
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.190370876713944e-05,
+      "loss": 0.129,
+      "step": 5658
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1898842037412958e-05,
+      "loss": 0.0936,
+      "step": 5659
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1893975648688426e-05,
+      "loss": 0.0881,
+      "step": 5660
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1889109601500956e-05,
+      "loss": 0.2016,
+      "step": 5661
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1884243896385622e-05,
+      "loss": 0.0927,
+      "step": 5662
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.187937853387744e-05,
+      "loss": 0.1358,
+      "step": 5663
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1874513514511414e-05,
+      "loss": 0.0624,
+      "step": 5664
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1869648838822496e-05,
+      "loss": 0.1272,
+      "step": 5665
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1864784507345605e-05,
+      "loss": 0.1457,
+      "step": 5666
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1859920520615622e-05,
+      "loss": 0.1101,
+      "step": 5667
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1855056879167382e-05,
+      "loss": 0.0741,
+      "step": 5668
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1850193583535698e-05,
+      "loss": 0.0818,
+      "step": 5669
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1845330634255327e-05,
+      "loss": 0.1075,
+      "step": 5670
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1840468031861009e-05,
+      "loss": 0.089,
+      "step": 5671
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.183560577688742e-05,
+      "loss": 0.1661,
+      "step": 5672
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1830743869869217e-05,
+      "loss": 0.0682,
+      "step": 5673
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1825882311341013e-05,
+      "loss": 0.1306,
+      "step": 5674
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1821021101837392e-05,
+      "loss": 0.1402,
+      "step": 5675
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.181616024189287e-05,
+      "loss": 0.0926,
+      "step": 5676
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1811299732041959e-05,
+      "loss": 0.0616,
+      "step": 5677
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1806439572819118e-05,
+      "loss": 0.121,
+      "step": 5678
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.180157976475877e-05,
+      "loss": 0.0556,
+      "step": 5679
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1796720308395289e-05,
+      "loss": 0.1179,
+      "step": 5680
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1791861204263023e-05,
+      "loss": 0.1142,
+      "step": 5681
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1787002452896275e-05,
+      "loss": 0.1385,
+      "step": 5682
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.178214405482932e-05,
+      "loss": 0.1198,
+      "step": 5683
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1777286010596375e-05,
+      "loss": 0.1005,
+      "step": 5684
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.177242832073163e-05,
+      "loss": 0.0888,
+      "step": 5685
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.176757098576924e-05,
+      "loss": 0.0959,
+      "step": 5686
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1762714006243312e-05,
+      "loss": 0.1216,
+      "step": 5687
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1757857382687912e-05,
+      "loss": 0.1733,
+      "step": 5688
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.175300111563708e-05,
+      "loss": 0.0794,
+      "step": 5689
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1748145205624805e-05,
+      "loss": 0.1312,
+      "step": 5690
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1743289653185046e-05,
+      "loss": 0.0825,
+      "step": 5691
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.173843445885171e-05,
+      "loss": 0.0899,
+      "step": 5692
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1733579623158673e-05,
+      "loss": 0.1175,
+      "step": 5693
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1728725146639775e-05,
+      "loss": 0.0927,
+      "step": 5694
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.172387102982881e-05,
+      "loss": 0.122,
+      "step": 5695
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1719017273259538e-05,
+      "loss": 0.1537,
+      "step": 5696
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.171416387746567e-05,
+      "loss": 0.1096,
+      "step": 5697
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1709310842980881e-05,
+      "loss": 0.1073,
+      "step": 5698
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1704458170338815e-05,
+      "loss": 0.1373,
+      "step": 5699
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1699605860073078e-05,
+      "loss": 0.1115,
+      "step": 5700
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1694753912717206e-05,
+      "loss": 0.1448,
+      "step": 5701
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1689902328804731e-05,
+      "loss": 0.0979,
+      "step": 5702
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1685051108869131e-05,
+      "loss": 0.0949,
+      "step": 5703
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1680200253443846e-05,
+      "loss": 0.1447,
+      "step": 5704
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1675349763062262e-05,
+      "loss": 0.0827,
+      "step": 5705
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1670499638257747e-05,
+      "loss": 0.1086,
+      "step": 5706
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1665649879563615e-05,
+      "loss": 0.0496,
+      "step": 5707
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.166080048751315e-05,
+      "loss": 0.0841,
+      "step": 5708
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.165595146263958e-05,
+      "loss": 0.0895,
+      "step": 5709
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1651102805476101e-05,
+      "loss": 0.1136,
+      "step": 5710
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1646254516555877e-05,
+      "loss": 0.0701,
+      "step": 5711
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1641406596412024e-05,
+      "loss": 0.1086,
+      "step": 5712
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1636559045577605e-05,
+      "loss": 0.1525,
+      "step": 5713
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1631711864585666e-05,
+      "loss": 0.1098,
+      "step": 5714
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1626865053969195e-05,
+      "loss": 0.0921,
+      "step": 5715
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1622018614261155e-05,
+      "loss": 0.1403,
+      "step": 5716
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1617172545994446e-05,
+      "loss": 0.1385,
+      "step": 5717
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1612326849701943e-05,
+      "loss": 0.1526,
+      "step": 5718
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.160748152591648e-05,
+      "loss": 0.1366,
+      "step": 5719
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1602636575170848e-05,
+      "loss": 0.1933,
+      "step": 5720
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1597791997997786e-05,
+      "loss": 0.1122,
+      "step": 5721
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1592947794930012e-05,
+      "loss": 0.1397,
+      "step": 5722
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1588103966500183e-05,
+      "loss": 0.0807,
+      "step": 5723
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.158326051324094e-05,
+      "loss": 0.0992,
+      "step": 5724
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.157841743568485e-05,
+      "loss": 0.0925,
+      "step": 5725
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.157357473436446e-05,
+      "loss": 0.1159,
+      "step": 5726
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1568732409812274e-05,
+      "loss": 0.077,
+      "step": 5727
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1563890462560753e-05,
+      "loss": 0.0819,
+      "step": 5728
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.155904889314232e-05,
+      "loss": 0.1572,
+      "step": 5729
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1554207702089336e-05,
+      "loss": 0.0967,
+      "step": 5730
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1549366889934152e-05,
+      "loss": 0.1061,
+      "step": 5731
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1544526457209048e-05,
+      "loss": 0.1045,
+      "step": 5732
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1539686404446293e-05,
+      "loss": 0.1068,
+      "step": 5733
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1534846732178083e-05,
+      "loss": 0.1021,
+      "step": 5734
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1530007440936588e-05,
+      "loss": 0.0825,
+      "step": 5735
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1525168531253938e-05,
+      "loss": 0.1264,
+      "step": 5736
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1520330003662222e-05,
+      "loss": 0.1038,
+      "step": 5737
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.151549185869347e-05,
+      "loss": 0.1071,
+      "step": 5738
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1510654096879692e-05,
+      "loss": 0.0972,
+      "step": 5739
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1505816718752838e-05,
+      "loss": 0.0744,
+      "step": 5740
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1500979724844834e-05,
+      "loss": 0.0826,
+      "step": 5741
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1496143115687549e-05,
+      "loss": 0.0758,
+      "step": 5742
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1491306891812809e-05,
+      "loss": 0.1013,
+      "step": 5743
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1486471053752413e-05,
+      "loss": 0.1279,
+      "step": 5744
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1481635602038103e-05,
+      "loss": 0.0809,
+      "step": 5745
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1476800537201577e-05,
+      "loss": 0.1262,
+      "step": 5746
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1471965859774507e-05,
+      "loss": 0.0895,
+      "step": 5747
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1467131570288503e-05,
+      "loss": 0.1517,
+      "step": 5748
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1462297669275152e-05,
+      "loss": 0.1353,
+      "step": 5749
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.145746415726598e-05,
+      "loss": 0.1235,
+      "step": 5750
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1452631034792474e-05,
+      "loss": 0.1196,
+      "step": 5751
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1447798302386089e-05,
+      "loss": 0.0642,
+      "step": 5752
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1442965960578235e-05,
+      "loss": 0.0916,
+      "step": 5753
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.143813400990026e-05,
+      "loss": 0.1255,
+      "step": 5754
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.143330245088349e-05,
+      "loss": 0.1322,
+      "step": 5755
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1428471284059205e-05,
+      "loss": 0.1583,
+      "step": 5756
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1423640509958632e-05,
+      "loss": 0.1075,
+      "step": 5757
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1418810129112968e-05,
+      "loss": 0.093,
+      "step": 5758
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1413980142053354e-05,
+      "loss": 0.1264,
+      "step": 5759
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1409150549310889e-05,
+      "loss": 0.1261,
+      "step": 5760
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1404321351416642e-05,
+      "loss": 0.095,
+      "step": 5761
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1399492548901629e-05,
+      "loss": 0.0908,
+      "step": 5762
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1394664142296813e-05,
+      "loss": 0.1148,
+      "step": 5763
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1389836132133132e-05,
+      "loss": 0.1109,
+      "step": 5764
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1385008518941468e-05,
+      "loss": 0.0683,
+      "step": 5765
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.138018130325267e-05,
+      "loss": 0.1789,
+      "step": 5766
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.137535448559753e-05,
+      "loss": 0.1216,
+      "step": 5767
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1370528066506798e-05,
+      "loss": 0.0826,
+      "step": 5768
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.13657020465112e-05,
+      "loss": 0.065,
+      "step": 5769
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1360876426141392e-05,
+      "loss": 0.1317,
+      "step": 5770
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1356051205927994e-05,
+      "loss": 0.1179,
+      "step": 5771
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1351226386401596e-05,
+      "loss": 0.1403,
+      "step": 5772
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.134640196809272e-05,
+      "loss": 0.1443,
+      "step": 5773
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1341577951531875e-05,
+      "loss": 0.1088,
+      "step": 5774
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1336754337249492e-05,
+      "loss": 0.0833,
+      "step": 5775
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1331931125775974e-05,
+      "loss": 0.1205,
+      "step": 5776
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1327108317641684e-05,
+      "loss": 0.1277,
+      "step": 5777
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1322285913376946e-05,
+      "loss": 0.0982,
+      "step": 5778
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.131746391351201e-05,
+      "loss": 0.1012,
+      "step": 5779
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.131264231857711e-05,
+      "loss": 0.0702,
+      "step": 5780
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1307821129102427e-05,
+      "loss": 0.1824,
+      "step": 5781
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1303000345618103e-05,
+      "loss": 0.0569,
+      "step": 5782
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1298179968654213e-05,
+      "loss": 0.0889,
+      "step": 5783
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1293359998740816e-05,
+      "loss": 0.1107,
+      "step": 5784
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1288540436407908e-05,
+      "loss": 0.1184,
+      "step": 5785
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1283721282185453e-05,
+      "loss": 0.1267,
+      "step": 5786
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.127890253660336e-05,
+      "loss": 0.108,
+      "step": 5787
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1274084200191491e-05,
+      "loss": 0.1222,
+      "step": 5788
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1269266273479674e-05,
+      "loss": 0.0956,
+      "step": 5789
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1264448756997678e-05,
+      "loss": 0.0776,
+      "step": 5790
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.125963165127525e-05,
+      "loss": 0.104,
+      "step": 5791
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1254814956842066e-05,
+      "loss": 0.1482,
+      "step": 5792
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1249998674227765e-05,
+      "loss": 0.0833,
+      "step": 5793
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.124518280396195e-05,
+      "loss": 0.0864,
+      "step": 5794
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1240367346574174e-05,
+      "loss": 0.0958,
+      "step": 5795
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1235552302593931e-05,
+      "loss": 0.0885,
+      "step": 5796
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1230737672550691e-05,
+      "loss": 0.1268,
+      "step": 5797
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1225923456973863e-05,
+      "loss": 0.0798,
+      "step": 5798
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1221109656392825e-05,
+      "loss": 0.0956,
+      "step": 5799
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1216296271336895e-05,
+      "loss": 0.0833,
+      "step": 5800
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1211483302335342e-05,
+      "loss": 0.1159,
+      "step": 5801
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.120667074991741e-05,
+      "loss": 0.0805,
+      "step": 5802
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1201858614612284e-05,
+      "loss": 0.1166,
+      "step": 5803
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1197046896949097e-05,
+      "loss": 0.1296,
+      "step": 5804
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1192235597456948e-05,
+      "loss": 0.1286,
+      "step": 5805
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1187424716664885e-05,
+      "loss": 0.1216,
+      "step": 5806
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1182614255101916e-05,
+      "loss": 0.0932,
+      "step": 5807
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1177804213296985e-05,
+      "loss": 0.1192,
+      "step": 5808
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1172994591779011e-05,
+      "loss": 0.0679,
+      "step": 5809
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1168185391076853e-05,
+      "loss": 0.0978,
+      "step": 5810
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1163376611719336e-05,
+      "loss": 0.1444,
+      "step": 5811
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1158568254235224e-05,
+      "loss": 0.0631,
+      "step": 5812
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1153760319153241e-05,
+      "loss": 0.0979,
+      "step": 5813
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1148952807002072e-05,
+      "loss": 0.171,
+      "step": 5814
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1144145718310347e-05,
+      "loss": 0.1494,
+      "step": 5815
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1139339053606647e-05,
+      "loss": 0.0896,
+      "step": 5816
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1134532813419513e-05,
+      "loss": 0.1134,
+      "step": 5817
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1129726998277435e-05,
+      "loss": 0.1218,
+      "step": 5818
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1124921608708865e-05,
+      "loss": 0.1099,
+      "step": 5819
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1120116645242202e-05,
+      "loss": 0.0888,
+      "step": 5820
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1115312108405785e-05,
+      "loss": 0.061,
+      "step": 5821
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.111050799872793e-05,
+      "loss": 0.126,
+      "step": 5822
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1105704316736888e-05,
+      "loss": 0.077,
+      "step": 5823
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.110090106296088e-05,
+      "loss": 0.0653,
+      "step": 5824
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.109609823792806e-05,
+      "loss": 0.123,
+      "step": 5825
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1091295842166544e-05,
+      "loss": 0.1343,
+      "step": 5826
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1086493876204408e-05,
+      "loss": 0.0909,
+      "step": 5827
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1081692340569674e-05,
+      "loss": 0.152,
+      "step": 5828
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1076891235790307e-05,
+      "loss": 0.0876,
+      "step": 5829
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1072090562394242e-05,
+      "loss": 0.1257,
+      "step": 5830
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.106729032090936e-05,
+      "loss": 0.1069,
+      "step": 5831
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1062490511863495e-05,
+      "loss": 0.1268,
+      "step": 5832
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1057691135784421e-05,
+      "loss": 0.0587,
+      "step": 5833
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1052892193199884e-05,
+      "loss": 0.1371,
+      "step": 5834
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1048093684637574e-05,
+      "loss": 0.1041,
+      "step": 5835
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1043295610625133e-05,
+      "loss": 0.0702,
+      "step": 5836
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.103849797169015e-05,
+      "loss": 0.1024,
+      "step": 5837
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1033700768360172e-05,
+      "loss": 0.0907,
+      "step": 5838
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1028904001162701e-05,
+      "loss": 0.1118,
+      "step": 5839
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1024107670625193e-05,
+      "loss": 0.1523,
+      "step": 5840
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1019311777275034e-05,
+      "loss": 0.0833,
+      "step": 5841
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.101451632163959e-05,
+      "loss": 0.0915,
+      "step": 5842
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1009721304246166e-05,
+      "loss": 0.1181,
+      "step": 5843
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1004926725622021e-05,
+      "loss": 0.1136,
+      "step": 5844
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1000132586294362e-05,
+      "loss": 0.0826,
+      "step": 5845
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0995338886790345e-05,
+      "loss": 0.0746,
+      "step": 5846
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0990545627637095e-05,
+      "loss": 0.0745,
+      "step": 5847
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0985752809361665e-05,
+      "loss": 0.11,
+      "step": 5848
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0980960432491083e-05,
+      "loss": 0.1278,
+      "step": 5849
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0976168497552307e-05,
+      "loss": 0.125,
+      "step": 5850
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.097137700507226e-05,
+      "loss": 0.1155,
+      "step": 5851
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0966585955577814e-05,
+      "loss": 0.1,
+      "step": 5852
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0961795349595788e-05,
+      "loss": 0.0718,
+      "step": 5853
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0957005187652954e-05,
+      "loss": 0.0832,
+      "step": 5854
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0952215470276036e-05,
+      "loss": 0.0834,
+      "step": 5855
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0947426197991714e-05,
+      "loss": 0.1058,
+      "step": 5856
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0942637371326615e-05,
+      "loss": 0.1848,
+      "step": 5857
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0937848990807303e-05,
+      "loss": 0.1006,
+      "step": 5858
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0933061056960322e-05,
+      "loss": 0.0684,
+      "step": 5859
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.092827357031214e-05,
+      "loss": 0.0883,
+      "step": 5860
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0923486531389196e-05,
+      "loss": 0.1516,
+      "step": 5861
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0918699940717866e-05,
+      "loss": 0.0808,
+      "step": 5862
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0913913798824477e-05,
+      "loss": 0.0883,
+      "step": 5863
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.090912810623532e-05,
+      "loss": 0.1331,
+      "step": 5864
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0904342863476625e-05,
+      "loss": 0.1429,
+      "step": 5865
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0899558071074568e-05,
+      "loss": 0.088,
+      "step": 5866
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0894773729555292e-05,
+      "loss": 0.1358,
+      "step": 5867
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0889989839444873e-05,
+      "loss": 0.0935,
+      "step": 5868
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0885206401269355e-05,
+      "loss": 0.1309,
+      "step": 5869
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0880423415554716e-05,
+      "loss": 0.0823,
+      "step": 5870
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0875640882826887e-05,
+      "loss": 0.1197,
+      "step": 5871
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0870858803611765e-05,
+      "loss": 0.1005,
+      "step": 5872
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.086607717843518e-05,
+      "loss": 0.1234,
+      "step": 5873
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0861296007822908e-05,
+      "loss": 0.0906,
+      "step": 5874
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0856515292300699e-05,
+      "loss": 0.0813,
+      "step": 5875
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0851735032394227e-05,
+      "loss": 0.117,
+      "step": 5876
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0846955228629134e-05,
+      "loss": 0.1278,
+      "step": 5877
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0842175881531008e-05,
+      "loss": 0.0736,
+      "step": 5878
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0837396991625374e-05,
+      "loss": 0.1509,
+      "step": 5879
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.083261855943772e-05,
+      "loss": 0.1455,
+      "step": 5880
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0827840585493487e-05,
+      "loss": 0.0927,
+      "step": 5881
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0823063070318055e-05,
+      "loss": 0.0498,
+      "step": 5882
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0818286014436753e-05,
+      "loss": 0.1139,
+      "step": 5883
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0813509418374869e-05,
+      "loss": 0.1081,
+      "step": 5884
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.080873328265763e-05,
+      "loss": 0.1367,
+      "step": 5885
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.080395760781023e-05,
+      "loss": 0.0743,
+      "step": 5886
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0799182394357787e-05,
+      "loss": 0.15,
+      "step": 5887
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0794407642825383e-05,
+      "loss": 0.1169,
+      "step": 5888
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0789633353738053e-05,
+      "loss": 0.0716,
+      "step": 5889
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0784859527620779e-05,
+      "loss": 0.1266,
+      "step": 5890
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0780086164998475e-05,
+      "loss": 0.1094,
+      "step": 5891
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0775313266396028e-05,
+      "loss": 0.1314,
+      "step": 5892
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.077054083233826e-05,
+      "loss": 0.0917,
+      "step": 5893
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0765768863349955e-05,
+      "loss": 0.1281,
+      "step": 5894
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0760997359955822e-05,
+      "loss": 0.1421,
+      "step": 5895
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0756226322680542e-05,
+      "loss": 0.1098,
+      "step": 5896
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0751455752048733e-05,
+      "loss": 0.1473,
+      "step": 5897
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0746685648584973e-05,
+      "loss": 0.1527,
+      "step": 5898
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0741916012813767e-05,
+      "loss": 0.1173,
+      "step": 5899
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0737146845259592e-05,
+      "loss": 0.0945,
+      "step": 5900
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0732378146446856e-05,
+      "loss": 0.0905,
+      "step": 5901
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0727609916899936e-05,
+      "loss": 0.1346,
+      "step": 5902
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.072284215714313e-05,
+      "loss": 0.1273,
+      "step": 5903
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0718074867700704e-05,
+      "loss": 0.1035,
+      "step": 5904
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.071330804909687e-05,
+      "loss": 0.0915,
+      "step": 5905
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.070854170185578e-05,
+      "loss": 0.1289,
+      "step": 5906
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.070377582650155e-05,
+      "loss": 0.1331,
+      "step": 5907
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0699010423558218e-05,
+      "loss": 0.0529,
+      "step": 5908
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0694245493549796e-05,
+      "loss": 0.0695,
+      "step": 5909
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0689481037000228e-05,
+      "loss": 0.1347,
+      "step": 5910
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0684717054433423e-05,
+      "loss": 0.1607,
+      "step": 5911
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0679953546373213e-05,
+      "loss": 0.1223,
+      "step": 5912
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0675190513343394e-05,
+      "loss": 0.1392,
+      "step": 5913
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.067042795586771e-05,
+      "loss": 0.1592,
+      "step": 5914
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0665665874469858e-05,
+      "loss": 0.0483,
+      "step": 5915
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0660904269673455e-05,
+      "loss": 0.1059,
+      "step": 5916
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0656143142002098e-05,
+      "loss": 0.0865,
+      "step": 5917
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0651382491979313e-05,
+      "loss": 0.1049,
+      "step": 5918
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.064662232012859e-05,
+      "loss": 0.1129,
+      "step": 5919
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0641862626973342e-05,
+      "loss": 0.1397,
+      "step": 5920
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0637103413036945e-05,
+      "loss": 0.0745,
+      "step": 5921
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0632344678842726e-05,
+      "loss": 0.077,
+      "step": 5922
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0627586424913954e-05,
+      "loss": 0.0965,
+      "step": 5923
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0622828651773834e-05,
+      "loss": 0.0872,
+      "step": 5924
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.061807135994554e-05,
+      "loss": 0.1148,
+      "step": 5925
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0613314549952172e-05,
+      "loss": 0.1096,
+      "step": 5926
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.06085582223168e-05,
+      "loss": 0.1218,
+      "step": 5927
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0603802377562416e-05,
+      "loss": 0.1359,
+      "step": 5928
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0599047016211968e-05,
+      "loss": 0.0778,
+      "step": 5929
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0594292138788367e-05,
+      "loss": 0.175,
+      "step": 5930
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.058953774581445e-05,
+      "loss": 0.1185,
+      "step": 5931
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0584783837813001e-05,
+      "loss": 0.1029,
+      "step": 5932
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0580030415306767e-05,
+      "loss": 0.0906,
+      "step": 5933
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.057527747881843e-05,
+      "loss": 0.0839,
+      "step": 5934
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0570525028870625e-05,
+      "loss": 0.132,
+      "step": 5935
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0565773065985919e-05,
+      "loss": 0.0947,
+      "step": 5936
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0561021590686842e-05,
+      "loss": 0.0975,
+      "step": 5937
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0556270603495861e-05,
+      "loss": 0.1165,
+      "step": 5938
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0551520104935399e-05,
+      "loss": 0.1146,
+      "step": 5939
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0546770095527817e-05,
+      "loss": 0.0732,
+      "step": 5940
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0542020575795416e-05,
+      "loss": 0.1064,
+      "step": 5941
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0537271546260457e-05,
+      "loss": 0.108,
+      "step": 5942
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0532523007445143e-05,
+      "loss": 0.129,
+      "step": 5943
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.052777495987162e-05,
+      "loss": 0.1506,
+      "step": 5944
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.052302740406198e-05,
+      "loss": 0.0693,
+      "step": 5945
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0518280340538257e-05,
+      "loss": 0.1106,
+      "step": 5946
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0513533769822447e-05,
+      "loss": 0.0939,
+      "step": 5947
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0508787692436476e-05,
+      "loss": 0.0762,
+      "step": 5948
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0504042108902214e-05,
+      "loss": 0.1027,
+      "step": 5949
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0499297019741495e-05,
+      "loss": 0.101,
+      "step": 5950
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0494552425476074e-05,
+      "loss": 0.1187,
+      "step": 5951
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0489808326627681e-05,
+      "loss": 0.1355,
+      "step": 5952
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.048506472371796e-05,
+      "loss": 0.1014,
+      "step": 5953
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0480321617268517e-05,
+      "loss": 0.0999,
+      "step": 5954
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.047557900780091e-05,
+      "loss": 0.1189,
+      "step": 5955
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0470836895836634e-05,
+      "loss": 0.087,
+      "step": 5956
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0466095281897118e-05,
+      "loss": 0.1208,
+      "step": 5957
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0461354166503755e-05,
+      "loss": 0.1054,
+      "step": 5958
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.045661355017788e-05,
+      "loss": 0.1191,
+      "step": 5959
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.045187343344077e-05,
+      "loss": 0.0937,
+      "step": 5960
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0447133816813633e-05,
+      "loss": 0.0746,
+      "step": 5961
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0442394700817646e-05,
+      "loss": 0.0502,
+      "step": 5962
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0437656085973915e-05,
+      "loss": 0.0998,
+      "step": 5963
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0432917972803506e-05,
+      "loss": 0.1576,
+      "step": 5964
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.042818036182741e-05,
+      "loss": 0.149,
+      "step": 5965
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0423443253566569e-05,
+      "loss": 0.0696,
+      "step": 5966
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0418706648541885e-05,
+      "loss": 0.121,
+      "step": 5967
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0413970547274184e-05,
+      "loss": 0.0762,
+      "step": 5968
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0409234950284254e-05,
+      "loss": 0.0892,
+      "step": 5969
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0404499858092815e-05,
+      "loss": 0.0714,
+      "step": 5970
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0399765271220527e-05,
+      "loss": 0.0954,
+      "step": 5971
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0395031190188015e-05,
+      "loss": 0.0778,
+      "step": 5972
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0390297615515837e-05,
+      "loss": 0.1391,
+      "step": 5973
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0385564547724481e-05,
+      "loss": 0.1092,
+      "step": 5974
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0380831987334408e-05,
+      "loss": 0.1195,
+      "step": 5975
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0376099934866e-05,
+      "loss": 0.1036,
+      "step": 5976
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0371368390839597e-05,
+      "loss": 0.1369,
+      "step": 5977
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.036663735577547e-05,
+      "loss": 0.1248,
+      "step": 5978
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0361906830193847e-05,
+      "loss": 0.0644,
+      "step": 5979
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0357176814614895e-05,
+      "loss": 0.0769,
+      "step": 5980
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0352447309558724e-05,
+      "loss": 0.0597,
+      "step": 5981
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0347718315545382e-05,
+      "loss": 0.0991,
+      "step": 5982
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0342989833094873e-05,
+      "loss": 0.0914,
+      "step": 5983
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0338261862727139e-05,
+      "loss": 0.1022,
+      "step": 5984
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0333534404962068e-05,
+      "loss": 0.1081,
+      "step": 5985
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.032880746031948e-05,
+      "loss": 0.0956,
+      "step": 5986
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0324081029319156e-05,
+      "loss": 0.1014,
+      "step": 5987
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0319355112480805e-05,
+      "loss": 0.1065,
+      "step": 5988
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0314629710324098e-05,
+      "loss": 0.1002,
+      "step": 5989
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.030990482336863e-05,
+      "loss": 0.1046,
+      "step": 5990
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0305180452133945e-05,
+      "loss": 0.1316,
+      "step": 5991
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.030045659713954e-05,
+      "loss": 0.1327,
+      "step": 5992
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0295733258904847e-05,
+      "loss": 0.1138,
+      "step": 5993
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0291010437949235e-05,
+      "loss": 0.1349,
+      "step": 5994
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.028628813479203e-05,
+      "loss": 0.1298,
+      "step": 5995
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0281566349952489e-05,
+      "loss": 0.14,
+      "step": 5996
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0276845083949823e-05,
+      "loss": 0.1391,
+      "step": 5997
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0272124337303185e-05,
+      "loss": 0.1332,
+      "step": 5998
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.026740411053165e-05,
+      "loss": 0.0548,
+      "step": 5999
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0262684404154265e-05,
+      "loss": 0.1079,
+      "step": 6000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.025796521869e-05,
+      "loss": 0.1097,
+      "step": 6001
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0253246554657785e-05,
+      "loss": 0.0839,
+      "step": 6002
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0248528412576472e-05,
+      "loss": 0.0918,
+      "step": 6003
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0243810792964865e-05,
+      "loss": 0.1249,
+      "step": 6004
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0239093696341718e-05,
+      "loss": 0.1062,
+      "step": 6005
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0234377123225721e-05,
+      "loss": 0.125,
+      "step": 6006
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0229661074135498e-05,
+      "loss": 0.0798,
+      "step": 6007
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.022494554958963e-05,
+      "loss": 0.0627,
+      "step": 6008
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0220230550106633e-05,
+      "loss": 0.1093,
+      "step": 6009
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0215516076204972e-05,
+      "loss": 0.1053,
+      "step": 6010
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0210802128403033e-05,
+      "loss": 0.0902,
+      "step": 6011
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0206088707219176e-05,
+      "loss": 0.1269,
+      "step": 6012
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0201375813171675e-05,
+      "loss": 0.0856,
+      "step": 6013
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0196663446778769e-05,
+      "loss": 0.1111,
+      "step": 6014
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.019195160855862e-05,
+      "loss": 0.1282,
+      "step": 6015
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0187240299029336e-05,
+      "loss": 0.0982,
+      "step": 6016
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0182529518708978e-05,
+      "loss": 0.1456,
+      "step": 6017
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0177819268115543e-05,
+      "loss": 0.0979,
+      "step": 6018
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.017310954776696e-05,
+      "loss": 0.1053,
+      "step": 6019
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0168400358181113e-05,
+      "loss": 0.0617,
+      "step": 6020
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0163691699875819e-05,
+      "loss": 0.1121,
+      "step": 6021
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0158983573368847e-05,
+      "loss": 0.1268,
+      "step": 6022
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0154275979177892e-05,
+      "loss": 0.1108,
+      "step": 6023
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0149568917820601e-05,
+      "loss": 0.1102,
+      "step": 6024
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0144862389814565e-05,
+      "loss": 0.0995,
+      "step": 6025
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.014015639567731e-05,
+      "loss": 0.1087,
+      "step": 6026
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0135450935926298e-05,
+      "loss": 0.0783,
+      "step": 6027
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.013074601107895e-05,
+      "loss": 0.1108,
+      "step": 6028
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0126041621652608e-05,
+      "loss": 0.1163,
+      "step": 6029
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0121337768164573e-05,
+      "loss": 0.1467,
+      "step": 6030
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0116634451132076e-05,
+      "loss": 0.1199,
+      "step": 6031
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0111931671072287e-05,
+      "loss": 0.1496,
+      "step": 6032
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0107229428502326e-05,
+      "loss": 0.0917,
+      "step": 6033
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0102527723939244e-05,
+      "loss": 0.1498,
+      "step": 6034
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0097826557900051e-05,
+      "loss": 0.0726,
+      "step": 6035
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.009312593090167e-05,
+      "loss": 0.1197,
+      "step": 6036
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0088425843460992e-05,
+      "loss": 0.1123,
+      "step": 6037
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0083726296094825e-05,
+      "loss": 0.1188,
+      "step": 6038
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0079027289319944e-05,
+      "loss": 0.1325,
+      "step": 6039
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0074328823653038e-05,
+      "loss": 0.0986,
+      "step": 6040
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0069630899610749e-05,
+      "loss": 0.081,
+      "step": 6041
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.006493351770966e-05,
+      "loss": 0.1137,
+      "step": 6042
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.00602366784663e-05,
+      "loss": 0.0695,
+      "step": 6043
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.005554038239712e-05,
+      "loss": 0.1365,
+      "step": 6044
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.005084463001853e-05,
+      "loss": 0.0789,
+      "step": 6045
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.004614942184687e-05,
+      "loss": 0.0778,
+      "step": 6046
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0041454758398425e-05,
+      "loss": 0.0604,
+      "step": 6047
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0036760640189418e-05,
+      "loss": 0.0954,
+      "step": 6048
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0032067067736007e-05,
+      "loss": 0.0992,
+      "step": 6049
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0027374041554304e-05,
+      "loss": 0.113,
+      "step": 6050
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0022681562160349e-05,
+      "loss": 0.0921,
+      "step": 6051
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0017989630070116e-05,
+      "loss": 0.094,
+      "step": 6052
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.001329824579954e-05,
+      "loss": 0.0864,
+      "step": 6053
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0008607409864473e-05,
+      "loss": 0.1085,
+      "step": 6054
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0003917122780733e-05,
+      "loss": 0.1858,
+      "step": 6055
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.999227385064045e-06,
+      "loss": 0.1346,
+      "step": 6056
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.994538197230098e-06,
+      "loss": 0.0977,
+      "step": 6057
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.989849559794513e-06,
+      "loss": 0.0996,
+      "step": 6058
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.985161473272846e-06,
+      "loss": 0.1213,
+      "step": 6059
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.980473938180607e-06,
+      "loss": 0.0643,
+      "step": 6060
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.975786955033226e-06,
+      "loss": 0.1257,
+      "step": 6061
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.971100524346085e-06,
+      "loss": 0.1252,
+      "step": 6062
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.966414646634498e-06,
+      "loss": 0.138,
+      "step": 6063
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.961729322413733e-06,
+      "loss": 0.1162,
+      "step": 6064
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.957044552198974e-06,
+      "loss": 0.0712,
+      "step": 6065
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.952360336505358e-06,
+      "loss": 0.1132,
+      "step": 6066
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.947676675847967e-06,
+      "loss": 0.1249,
+      "step": 6067
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.942993570741812e-06,
+      "loss": 0.1279,
+      "step": 6068
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.938311021701834e-06,
+      "loss": 0.0571,
+      "step": 6069
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.933629029242939e-06,
+      "loss": 0.1504,
+      "step": 6070
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.928947593879945e-06,
+      "loss": 0.1558,
+      "step": 6071
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.924266716127633e-06,
+      "loss": 0.0604,
+      "step": 6072
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.9195863965007e-06,
+      "loss": 0.129,
+      "step": 6073
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.914906635513793e-06,
+      "loss": 0.0955,
+      "step": 6074
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.9102274336815e-06,
+      "loss": 0.086,
+      "step": 6075
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.905548791518347e-06,
+      "loss": 0.0569,
+      "step": 6076
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.900870709538783e-06,
+      "loss": 0.1301,
+      "step": 6077
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.89619318825722e-06,
+      "loss": 0.1087,
+      "step": 6078
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.89151622818799e-06,
+      "loss": 0.1441,
+      "step": 6079
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.886839829845375e-06,
+      "loss": 0.0353,
+      "step": 6080
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.882163993743584e-06,
+      "loss": 0.0636,
+      "step": 6081
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.87748872039677e-06,
+      "loss": 0.0994,
+      "step": 6082
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.872814010319028e-06,
+      "loss": 0.0791,
+      "step": 6083
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.868139864024389e-06,
+      "loss": 0.096,
+      "step": 6084
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.863466282026806e-06,
+      "loss": 0.0833,
+      "step": 6085
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.858793264840201e-06,
+      "loss": 0.1349,
+      "step": 6086
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.854120812978405e-06,
+      "loss": 0.0729,
+      "step": 6087
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.849448926955204e-06,
+      "loss": 0.1801,
+      "step": 6088
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.844777607284322e-06,
+      "loss": 0.099,
+      "step": 6089
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.840106854479408e-06,
+      "loss": 0.1084,
+      "step": 6090
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.835436669054052e-06,
+      "loss": 0.1496,
+      "step": 6091
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.830767051521796e-06,
+      "loss": 0.1198,
+      "step": 6092
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.826098002396106e-06,
+      "loss": 0.1182,
+      "step": 6093
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.821429522190379e-06,
+      "loss": 0.1288,
+      "step": 6094
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.81676161141797e-06,
+      "loss": 0.0789,
+      "step": 6095
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.812094270592155e-06,
+      "loss": 0.1031,
+      "step": 6096
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.807427500226162e-06,
+      "loss": 0.0919,
+      "step": 6097
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.802761300833133e-06,
+      "loss": 0.1404,
+      "step": 6098
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.798095672926169e-06,
+      "loss": 0.1356,
+      "step": 6099
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.793430617018298e-06,
+      "loss": 0.1151,
+      "step": 6100
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.788766133622494e-06,
+      "loss": 0.1396,
+      "step": 6101
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.78410222325165e-06,
+      "loss": 0.1261,
+      "step": 6102
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.779438886418616e-06,
+      "loss": 0.1027,
+      "step": 6103
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.774776123636166e-06,
+      "loss": 0.1175,
+      "step": 6104
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.770113935417022e-06,
+      "loss": 0.1081,
+      "step": 6105
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.765452322273831e-06,
+      "loss": 0.0787,
+      "step": 6106
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.760791284719178e-06,
+      "loss": 0.0815,
+      "step": 6107
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.756130823265596e-06,
+      "loss": 0.0886,
+      "step": 6108
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.751470938425548e-06,
+      "loss": 0.0676,
+      "step": 6109
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.746811630711423e-06,
+      "loss": 0.0944,
+      "step": 6110
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.742152900635566e-06,
+      "loss": 0.1288,
+      "step": 6111
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.737494748710243e-06,
+      "loss": 0.106,
+      "step": 6112
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.73283717544767e-06,
+      "loss": 0.0669,
+      "step": 6113
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.728180181359983e-06,
+      "loss": 0.1522,
+      "step": 6114
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.723523766959268e-06,
+      "loss": 0.13,
+      "step": 6115
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.718867932757538e-06,
+      "loss": 0.1067,
+      "step": 6116
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.714212679266754e-06,
+      "loss": 0.1011,
+      "step": 6117
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.709558006998799e-06,
+      "loss": 0.0373,
+      "step": 6118
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.704903916465499e-06,
+      "loss": 0.0537,
+      "step": 6119
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.700250408178619e-06,
+      "loss": 0.0922,
+      "step": 6120
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.695597482649852e-06,
+      "loss": 0.0891,
+      "step": 6121
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.690945140390841e-06,
+      "loss": 0.167,
+      "step": 6122
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.686293381913148e-06,
+      "loss": 0.1526,
+      "step": 6123
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.681642207728275e-06,
+      "loss": 0.0956,
+      "step": 6124
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.676991618347671e-06,
+      "loss": 0.1061,
+      "step": 6125
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.672341614282713e-06,
+      "loss": 0.1061,
+      "step": 6126
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.6676921960447e-06,
+      "loss": 0.1188,
+      "step": 6127
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.663043364144897e-06,
+      "loss": 0.1658,
+      "step": 6128
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.658395119094476e-06,
+      "loss": 0.083,
+      "step": 6129
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.653747461404566e-06,
+      "loss": 0.1165,
+      "step": 6130
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.649100391586212e-06,
+      "loss": 0.124,
+      "step": 6131
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.644453910150406e-06,
+      "loss": 0.0975,
+      "step": 6132
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.639808017608076e-06,
+      "loss": 0.0689,
+      "step": 6133
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.635162714470082e-06,
+      "loss": 0.1284,
+      "step": 6134
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.630518001247215e-06,
+      "loss": 0.1213,
+      "step": 6135
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.62587387845021e-06,
+      "loss": 0.1144,
+      "step": 6136
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.621230346589728e-06,
+      "loss": 0.1224,
+      "step": 6137
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.616587406176383e-06,
+      "loss": 0.0816,
+      "step": 6138
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.611945057720691e-06,
+      "loss": 0.0803,
+      "step": 6139
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.607303301733134e-06,
+      "loss": 0.1534,
+      "step": 6140
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.602662138724115e-06,
+      "loss": 0.0583,
+      "step": 6141
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.598021569203982e-06,
+      "loss": 0.1039,
+      "step": 6142
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.593381593682999e-06,
+      "loss": 0.1789,
+      "step": 6143
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.588742212671376e-06,
+      "loss": 0.1227,
+      "step": 6144
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.584103426679264e-06,
+      "loss": 0.06,
+      "step": 6145
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.579465236216743e-06,
+      "loss": 0.0355,
+      "step": 6146
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.574827641793816e-06,
+      "loss": 0.079,
+      "step": 6147
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.57019064392044e-06,
+      "loss": 0.1104,
+      "step": 6148
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.565554243106489e-06,
+      "loss": 0.1143,
+      "step": 6149
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.56091843986179e-06,
+      "loss": 0.1074,
+      "step": 6150
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.556283234696091e-06,
+      "loss": 0.0919,
+      "step": 6151
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.551648628119066e-06,
+      "loss": 0.1343,
+      "step": 6152
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.54701462064035e-06,
+      "loss": 0.107,
+      "step": 6153
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.542381212769484e-06,
+      "loss": 0.0878,
+      "step": 6154
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.53774840501597e-06,
+      "loss": 0.1533,
+      "step": 6155
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.533116197889213e-06,
+      "loss": 0.1034,
+      "step": 6156
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.528484591898575e-06,
+      "loss": 0.087,
+      "step": 6157
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.52385358755335e-06,
+      "loss": 0.0714,
+      "step": 6158
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.519223185362759e-06,
+      "loss": 0.0943,
+      "step": 6159
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.514593385835952e-06,
+      "loss": 0.1117,
+      "step": 6160
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.509964189482028e-06,
+      "loss": 0.065,
+      "step": 6161
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.505335596810005e-06,
+      "loss": 0.117,
+      "step": 6162
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.500707608328852e-06,
+      "loss": 0.0946,
+      "step": 6163
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.496080224547449e-06,
+      "loss": 0.1134,
+      "step": 6164
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.491453445974619e-06,
+      "loss": 0.1417,
+      "step": 6165
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.486827273119129e-06,
+      "loss": 0.1088,
+      "step": 6166
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.482201706489676e-06,
+      "loss": 0.1461,
+      "step": 6167
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.477576746594875e-06,
+      "loss": 0.0598,
+      "step": 6168
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.472952393943283e-06,
+      "loss": 0.0692,
+      "step": 6169
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.468328649043397e-06,
+      "loss": 0.1097,
+      "step": 6170
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.463705512403646e-06,
+      "loss": 0.0753,
+      "step": 6171
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.459082984532376e-06,
+      "loss": 0.0932,
+      "step": 6172
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.45446106593789e-06,
+      "loss": 0.1019,
+      "step": 6173
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.449839757128403e-06,
+      "loss": 0.0803,
+      "step": 6174
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.445219058612082e-06,
+      "loss": 0.0753,
+      "step": 6175
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.440598970897011e-06,
+      "loss": 0.0572,
+      "step": 6176
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.435979494491206e-06,
+      "loss": 0.0883,
+      "step": 6177
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.431360629902634e-06,
+      "loss": 0.1386,
+      "step": 6178
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.426742377639178e-06,
+      "loss": 0.1369,
+      "step": 6179
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.422124738208665e-06,
+      "loss": 0.0418,
+      "step": 6180
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.417507712118839e-06,
+      "loss": 0.1079,
+      "step": 6181
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.41289129987739e-06,
+      "loss": 0.0863,
+      "step": 6182
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.40827550199194e-06,
+      "loss": 0.1092,
+      "step": 6183
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.40366031897004e-06,
+      "loss": 0.1244,
+      "step": 6184
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.399045751319163e-06,
+      "loss": 0.1059,
+      "step": 6185
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.394431799546736e-06,
+      "loss": 0.0799,
+      "step": 6186
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.389818464160103e-06,
+      "loss": 0.079,
+      "step": 6187
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.38520574566655e-06,
+      "loss": 0.1175,
+      "step": 6188
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.380593644573283e-06,
+      "loss": 0.1078,
+      "step": 6189
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.375982161387443e-06,
+      "loss": 0.0724,
+      "step": 6190
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.371371296616114e-06,
+      "loss": 0.094,
+      "step": 6191
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.366761050766305e-06,
+      "loss": 0.0522,
+      "step": 6192
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.362151424344954e-06,
+      "loss": 0.1416,
+      "step": 6193
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.35754241785893e-06,
+      "loss": 0.063,
+      "step": 6194
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.352934031815043e-06,
+      "loss": 0.121,
+      "step": 6195
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.34832626672003e-06,
+      "loss": 0.1262,
+      "step": 6196
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.34371912308055e-06,
+      "loss": 0.0405,
+      "step": 6197
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.339112601403213e-06,
+      "loss": 0.1187,
+      "step": 6198
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.334506702194542e-06,
+      "loss": 0.0949,
+      "step": 6199
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.329901425961007e-06,
+      "loss": 0.0747,
+      "step": 6200
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.325296773208997e-06,
+      "loss": 0.1498,
+      "step": 6201
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.320692744444835e-06,
+      "loss": 0.118,
+      "step": 6202
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.316089340174781e-06,
+      "loss": 0.1081,
+      "step": 6203
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.311486560905032e-06,
+      "loss": 0.0988,
+      "step": 6204
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.30688440714169e-06,
+      "loss": 0.1082,
+      "step": 6205
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.30228287939082e-06,
+      "loss": 0.0588,
+      "step": 6206
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.297681978158398e-06,
+      "loss": 0.1401,
+      "step": 6207
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.293081703950338e-06,
+      "loss": 0.1125,
+      "step": 6208
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.288482057272489e-06,
+      "loss": 0.2079,
+      "step": 6209
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.283883038630616e-06,
+      "loss": 0.1118,
+      "step": 6210
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.279284648530433e-06,
+      "loss": 0.0788,
+      "step": 6211
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.27468688747757e-06,
+      "loss": 0.0867,
+      "step": 6212
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.270089755977608e-06,
+      "loss": 0.0687,
+      "step": 6213
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.265493254536032e-06,
+      "loss": 0.1185,
+      "step": 6214
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.260897383658273e-06,
+      "loss": 0.0756,
+      "step": 6215
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.256302143849693e-06,
+      "loss": 0.1295,
+      "step": 6216
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.251707535615591e-06,
+      "loss": 0.0925,
+      "step": 6217
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.24711355946118e-06,
+      "loss": 0.1472,
+      "step": 6218
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.242520215891605e-06,
+      "loss": 0.1675,
+      "step": 6219
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.23792750541196e-06,
+      "loss": 0.1357,
+      "step": 6220
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.233335428527257e-06,
+      "loss": 0.1523,
+      "step": 6221
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.228743985742425e-06,
+      "loss": 0.0657,
+      "step": 6222
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.224153177562352e-06,
+      "loss": 0.0835,
+      "step": 6223
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.21956300449183e-06,
+      "loss": 0.1909,
+      "step": 6224
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.214973467035607e-06,
+      "loss": 0.1712,
+      "step": 6225
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.210384565698334e-06,
+      "loss": 0.133,
+      "step": 6226
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.205796300984604e-06,
+      "loss": 0.1075,
+      "step": 6227
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.201208673398948e-06,
+      "loss": 0.0489,
+      "step": 6228
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.19662168344582e-06,
+      "loss": 0.0981,
+      "step": 6229
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.192035331629593e-06,
+      "loss": 0.0912,
+      "step": 6230
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.187449618454588e-06,
+      "loss": 0.085,
+      "step": 6231
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.182864544425044e-06,
+      "loss": 0.0952,
+      "step": 6232
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.178280110045143e-06,
+      "loss": 0.088,
+      "step": 6233
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.17369631581898e-06,
+      "loss": 0.1137,
+      "step": 6234
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.169113162250583e-06,
+      "loss": 0.1227,
+      "step": 6235
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.16453064984392e-06,
+      "loss": 0.0431,
+      "step": 6236
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.159948779102885e-06,
+      "loss": 0.0929,
+      "step": 6237
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.155367550531289e-06,
+      "loss": 0.1035,
+      "step": 6238
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.150786964632886e-06,
+      "loss": 0.1252,
+      "step": 6239
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.146207021911356e-06,
+      "loss": 0.1106,
+      "step": 6240
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.141627722870305e-06,
+      "loss": 0.0955,
+      "step": 6241
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.137049068013284e-06,
+      "loss": 0.1411,
+      "step": 6242
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.13247105784374e-06,
+      "loss": 0.0874,
+      "step": 6243
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.127893692865075e-06,
+      "loss": 0.1367,
+      "step": 6244
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.123316973580623e-06,
+      "loss": 0.1329,
+      "step": 6245
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.118740900493634e-06,
+      "loss": 0.1077,
+      "step": 6246
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.114165474107282e-06,
+      "loss": 0.1284,
+      "step": 6247
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.10959069492469e-06,
+      "loss": 0.0696,
+      "step": 6248
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.105016563448889e-06,
+      "loss": 0.0626,
+      "step": 6249
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.100443080182862e-06,
+      "loss": 0.1113,
+      "step": 6250
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.095870245629495e-06,
+      "loss": 0.1367,
+      "step": 6251
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.091298060291617e-06,
+      "loss": 0.1085,
+      "step": 6252
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.08672652467199e-06,
+      "loss": 0.0951,
+      "step": 6253
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.082155639273293e-06,
+      "loss": 0.053,
+      "step": 6254
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.077585404598135e-06,
+      "loss": 0.0428,
+      "step": 6255
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.073015821149064e-06,
+      "loss": 0.133,
+      "step": 6256
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.068446889428542e-06,
+      "loss": 0.1461,
+      "step": 6257
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.063878609938979e-06,
+      "loss": 0.091,
+      "step": 6258
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.059310983182688e-06,
+      "loss": 0.1138,
+      "step": 6259
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.054744009661926e-06,
+      "loss": 0.1328,
+      "step": 6260
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.05017768987888e-06,
+      "loss": 0.0476,
+      "step": 6261
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.045612024335661e-06,
+      "loss": 0.1079,
+      "step": 6262
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.0410470135343e-06,
+      "loss": 0.0851,
+      "step": 6263
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.03648265797677e-06,
+      "loss": 0.0998,
+      "step": 6264
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.031918958164962e-06,
+      "loss": 0.0764,
+      "step": 6265
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.027355914600707e-06,
+      "loss": 0.0677,
+      "step": 6266
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.022793527785744e-06,
+      "loss": 0.0661,
+      "step": 6267
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.018231798221751e-06,
+      "loss": 0.1085,
+      "step": 6268
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.013670726410339e-06,
+      "loss": 0.1081,
+      "step": 6269
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.009110312853046e-06,
+      "loss": 0.0755,
+      "step": 6270
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.004550558051328e-06,
+      "loss": 0.1119,
+      "step": 6271
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.99999146250657e-06,
+      "loss": 0.1035,
+      "step": 6272
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.995433026720092e-06,
+      "loss": 0.1355,
+      "step": 6273
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.990875251193135e-06,
+      "loss": 0.1099,
+      "step": 6274
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.986318136426877e-06,
+      "loss": 0.1195,
+      "step": 6275
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.981761682922409e-06,
+      "loss": 0.1985,
+      "step": 6276
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.977205891180755e-06,
+      "loss": 0.0985,
+      "step": 6277
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.972650761702876e-06,
+      "loss": 0.092,
+      "step": 6278
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.968096294989651e-06,
+      "loss": 0.0658,
+      "step": 6279
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.963542491541877e-06,
+      "loss": 0.078,
+      "step": 6280
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.9589893518603e-06,
+      "loss": 0.0637,
+      "step": 6281
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.954436876445572e-06,
+      "loss": 0.1126,
+      "step": 6282
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.949885065798291e-06,
+      "loss": 0.0896,
+      "step": 6283
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.945333920418963e-06,
+      "loss": 0.1084,
+      "step": 6284
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.940783440808032e-06,
+      "loss": 0.1141,
+      "step": 6285
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.936233627465873e-06,
+      "loss": 0.1113,
+      "step": 6286
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.931684480892775e-06,
+      "loss": 0.0623,
+      "step": 6287
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.927136001588959e-06,
+      "loss": 0.1304,
+      "step": 6288
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.92258819005458e-06,
+      "loss": 0.083,
+      "step": 6289
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.918041046789704e-06,
+      "loss": 0.0874,
+      "step": 6290
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.913494572294347e-06,
+      "loss": 0.1296,
+      "step": 6291
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.908948767068427e-06,
+      "loss": 0.0718,
+      "step": 6292
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.904403631611797e-06,
+      "loss": 0.0742,
+      "step": 6293
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.899859166424241e-06,
+      "loss": 0.0583,
+      "step": 6294
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.895315372005473e-06,
+      "loss": 0.095,
+      "step": 6295
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.89077224885512e-06,
+      "loss": 0.1207,
+      "step": 6296
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.88622979747274e-06,
+      "loss": 0.0484,
+      "step": 6297
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.881688018357824e-06,
+      "loss": 0.0713,
+      "step": 6298
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.877146912009777e-06,
+      "loss": 0.1147,
+      "step": 6299
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.872606478927951e-06,
+      "loss": 0.0784,
+      "step": 6300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.868066719611595e-06,
+      "loss": 0.1188,
+      "step": 6301
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.863527634559902e-06,
+      "loss": 0.0863,
+      "step": 6302
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.858989224271994e-06,
+      "loss": 0.1198,
+      "step": 6303
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.854451489246912e-06,
+      "loss": 0.1264,
+      "step": 6304
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.849914429983614e-06,
+      "loss": 0.108,
+      "step": 6305
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.845378046981002e-06,
+      "loss": 0.1456,
+      "step": 6306
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.84084234073789e-06,
+      "loss": 0.0986,
+      "step": 6307
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.836307311753032e-06,
+      "loss": 0.1555,
+      "step": 6308
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.831772960525083e-06,
+      "loss": 0.0919,
+      "step": 6309
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.827239287552646e-06,
+      "loss": 0.1675,
+      "step": 6310
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.822706293334243e-06,
+      "loss": 0.0955,
+      "step": 6311
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.81817397836832e-06,
+      "loss": 0.1478,
+      "step": 6312
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.813642343153242e-06,
+      "loss": 0.0915,
+      "step": 6313
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.80911138818731e-06,
+      "loss": 0.1099,
+      "step": 6314
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.804581113968747e-06,
+      "loss": 0.1014,
+      "step": 6315
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.800051520995703e-06,
+      "loss": 0.114,
+      "step": 6316
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.795522609766241e-06,
+      "loss": 0.0999,
+      "step": 6317
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.790994380778364e-06,
+      "loss": 0.1294,
+      "step": 6318
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.786466834529989e-06,
+      "loss": 0.1158,
+      "step": 6319
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.781939971518974e-06,
+      "loss": 0.1112,
+      "step": 6320
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.777413792243079e-06,
+      "loss": 0.0834,
+      "step": 6321
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.772888297200006e-06,
+      "loss": 0.1034,
+      "step": 6322
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.768363486887377e-06,
+      "loss": 0.0858,
+      "step": 6323
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.76383936180274e-06,
+      "loss": 0.1198,
+      "step": 6324
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.759315922443557e-06,
+      "loss": 0.0734,
+      "step": 6325
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.754793169307233e-06,
+      "loss": 0.0782,
+      "step": 6326
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.750271102891078e-06,
+      "loss": 0.0867,
+      "step": 6327
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.74574972369235e-06,
+      "loss": 0.0919,
+      "step": 6328
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.741229032208205e-06,
+      "loss": 0.0848,
+      "step": 6329
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.736709028935748e-06,
+      "loss": 0.1203,
+      "step": 6330
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.732189714371984e-06,
+      "loss": 0.0947,
+      "step": 6331
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.72767108901386e-06,
+      "loss": 0.0946,
+      "step": 6332
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.72315315335825e-06,
+      "loss": 0.1781,
+      "step": 6333
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.718635907901931e-06,
+      "loss": 0.1005,
+      "step": 6334
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.71411935314163e-06,
+      "loss": 0.0813,
+      "step": 6335
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.709603489573973e-06,
+      "loss": 0.0981,
+      "step": 6336
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.705088317695534e-06,
+      "loss": 0.0758,
+      "step": 6337
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.700573838002787e-06,
+      "loss": 0.1138,
+      "step": 6338
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.696060050992148e-06,
+      "loss": 0.0795,
+      "step": 6339
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.691546957159958e-06,
+      "loss": 0.1645,
+      "step": 6340
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.687034557002469e-06,
+      "loss": 0.074,
+      "step": 6341
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.682522851015856e-06,
+      "loss": 0.1041,
+      "step": 6342
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.67801183969623e-06,
+      "loss": 0.1902,
+      "step": 6343
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.673501523539622e-06,
+      "loss": 0.1116,
+      "step": 6344
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.668991903041986e-06,
+      "loss": 0.1488,
+      "step": 6345
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.66448297869919e-06,
+      "loss": 0.1303,
+      "step": 6346
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.65997475100704e-06,
+      "loss": 0.0904,
+      "step": 6347
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.655467220461255e-06,
+      "loss": 0.0874,
+      "step": 6348
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.650960387557485e-06,
+      "loss": 0.0935,
+      "step": 6349
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.646454252791293e-06,
+      "loss": 0.0804,
+      "step": 6350
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.641948816658174e-06,
+      "loss": 0.0764,
+      "step": 6351
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.637444079653552e-06,
+      "loss": 0.1018,
+      "step": 6352
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.63294004227276e-06,
+      "loss": 0.0974,
+      "step": 6353
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.628436705011049e-06,
+      "loss": 0.0611,
+      "step": 6354
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.623934068363616e-06,
+      "loss": 0.0812,
+      "step": 6355
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.619432132825568e-06,
+      "loss": 0.1212,
+      "step": 6356
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.614930898891938e-06,
+      "loss": 0.1135,
+      "step": 6357
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.610430367057679e-06,
+      "loss": 0.1227,
+      "step": 6358
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.60593053781766e-06,
+      "loss": 0.0954,
+      "step": 6359
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.601431411666683e-06,
+      "loss": 0.0415,
+      "step": 6360
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.596932989099471e-06,
+      "loss": 0.1247,
+      "step": 6361
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.592435270610676e-06,
+      "loss": 0.1003,
+      "step": 6362
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.587938256694858e-06,
+      "loss": 0.074,
+      "step": 6363
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.583441947846502e-06,
+      "loss": 0.0561,
+      "step": 6364
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.578946344560027e-06,
+      "loss": 0.0759,
+      "step": 6365
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.574451447329772e-06,
+      "loss": 0.045,
+      "step": 6366
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.56995725664998e-06,
+      "loss": 0.0951,
+      "step": 6367
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.565463773014844e-06,
+      "loss": 0.0664,
+      "step": 6368
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.560970996918456e-06,
+      "loss": 0.0836,
+      "step": 6369
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.556478928854846e-06,
+      "loss": 0.1401,
+      "step": 6370
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.551987569317954e-06,
+      "loss": 0.126,
+      "step": 6371
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.54749691880165e-06,
+      "loss": 0.0969,
+      "step": 6372
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.543006977799724e-06,
+      "loss": 0.1236,
+      "step": 6373
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.538517746805896e-06,
+      "loss": 0.1138,
+      "step": 6374
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.534029226313793e-06,
+      "loss": 0.075,
+      "step": 6375
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.529541416816962e-06,
+      "loss": 0.0516,
+      "step": 6376
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.525054318808891e-06,
+      "loss": 0.1489,
+      "step": 6377
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.520567932782984e-06,
+      "loss": 0.1607,
+      "step": 6378
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.516082259232547e-06,
+      "loss": 0.0961,
+      "step": 6379
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.511597298650839e-06,
+      "loss": 0.1487,
+      "step": 6380
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.507113051531007e-06,
+      "loss": 0.1151,
+      "step": 6381
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.502629518366156e-06,
+      "loss": 0.089,
+      "step": 6382
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.498146699649275e-06,
+      "loss": 0.0903,
+      "step": 6383
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.493664595873302e-06,
+      "loss": 0.0879,
+      "step": 6384
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.48918320753109e-06,
+      "loss": 0.0563,
+      "step": 6385
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.484702535115411e-06,
+      "loss": 0.1425,
+      "step": 6386
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.480222579118944e-06,
+      "loss": 0.1099,
+      "step": 6387
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.475743340034314e-06,
+      "loss": 0.084,
+      "step": 6388
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.471264818354056e-06,
+      "loss": 0.0952,
+      "step": 6389
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.466787014570628e-06,
+      "loss": 0.0977,
+      "step": 6390
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.462309929176403e-06,
+      "loss": 0.1469,
+      "step": 6391
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.457833562663678e-06,
+      "loss": 0.0858,
+      "step": 6392
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.453357915524675e-06,
+      "loss": 0.0732,
+      "step": 6393
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.448882988251531e-06,
+      "loss": 0.08,
+      "step": 6394
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.44440878133632e-06,
+      "loss": 0.0827,
+      "step": 6395
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.439935295271006e-06,
+      "loss": 0.0865,
+      "step": 6396
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.435462530547503e-06,
+      "loss": 0.1317,
+      "step": 6397
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.430990487657627e-06,
+      "loss": 0.1059,
+      "step": 6398
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.42651916709313e-06,
+      "loss": 0.1031,
+      "step": 6399
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.422048569345667e-06,
+      "loss": 0.0907,
+      "step": 6400
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.417578694906828e-06,
+      "loss": 0.1024,
+      "step": 6401
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.413109544268124e-06,
+      "loss": 0.0657,
+      "step": 6402
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.408641117920976e-06,
+      "loss": 0.0927,
+      "step": 6403
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.404173416356722e-06,
+      "loss": 0.1307,
+      "step": 6404
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.399706440066635e-06,
+      "loss": 0.0687,
+      "step": 6405
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.395240189541904e-06,
+      "loss": 0.0819,
+      "step": 6406
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.39077466527364e-06,
+      "loss": 0.0838,
+      "step": 6407
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.386309867752863e-06,
+      "loss": 0.1301,
+      "step": 6408
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.381845797470518e-06,
+      "loss": 0.1038,
+      "step": 6409
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.377382454917472e-06,
+      "loss": 0.0665,
+      "step": 6410
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.372919840584525e-06,
+      "loss": 0.0806,
+      "step": 6411
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.36845795496237e-06,
+      "loss": 0.0841,
+      "step": 6412
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.363996798541643e-06,
+      "loss": 0.1111,
+      "step": 6413
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.35953637181288e-06,
+      "loss": 0.1064,
+      "step": 6414
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.355076675266561e-06,
+      "loss": 0.0779,
+      "step": 6415
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.350617709393062e-06,
+      "loss": 0.0925,
+      "step": 6416
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.346159474682689e-06,
+      "loss": 0.078,
+      "step": 6417
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.34170197162568e-06,
+      "loss": 0.1339,
+      "step": 6418
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.337245200712163e-06,
+      "loss": 0.1004,
+      "step": 6419
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.332789162432218e-06,
+      "loss": 0.1112,
+      "step": 6420
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.328333857275815e-06,
+      "loss": 0.0954,
+      "step": 6421
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.323879285732864e-06,
+      "loss": 0.1079,
+      "step": 6422
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.319425448293189e-06,
+      "loss": 0.133,
+      "step": 6423
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.314972345446535e-06,
+      "loss": 0.1428,
+      "step": 6424
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.31051997768256e-06,
+      "loss": 0.0702,
+      "step": 6425
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.306068345490835e-06,
+      "loss": 0.0845,
+      "step": 6426
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.30161744936087e-06,
+      "loss": 0.0707,
+      "step": 6427
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.297167289782088e-06,
+      "loss": 0.1335,
+      "step": 6428
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.292717867243812e-06,
+      "loss": 0.0957,
+      "step": 6429
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.288269182235314e-06,
+      "loss": 0.1126,
+      "step": 6430
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.283821235245755e-06,
+      "loss": 0.1218,
+      "step": 6431
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.279374026764242e-06,
+      "loss": 0.1273,
+      "step": 6432
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.274927557279778e-06,
+      "loss": 0.177,
+      "step": 6433
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.270481827281297e-06,
+      "loss": 0.0645,
+      "step": 6434
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.26603683725766e-06,
+      "loss": 0.1483,
+      "step": 6435
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.261592587697626e-06,
+      "loss": 0.1118,
+      "step": 6436
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.257149079089879e-06,
+      "loss": 0.1249,
+      "step": 6437
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.25270631192303e-06,
+      "loss": 0.1425,
+      "step": 6438
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.248264286685604e-06,
+      "loss": 0.1043,
+      "step": 6439
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.243823003866052e-06,
+      "loss": 0.1397,
+      "step": 6440
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.239382463952727e-06,
+      "loss": 0.102,
+      "step": 6441
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.234942667433903e-06,
+      "loss": 0.0836,
+      "step": 6442
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.230503614797786e-06,
+      "loss": 0.0937,
+      "step": 6443
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.226065306532497e-06,
+      "loss": 0.0881,
+      "step": 6444
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.221627743126059e-06,
+      "loss": 0.0947,
+      "step": 6445
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.217190925066434e-06,
+      "loss": 0.1369,
+      "step": 6446
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.212754852841485e-06,
+      "loss": 0.0709,
+      "step": 6447
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.20831952693901e-06,
+      "loss": 0.0902,
+      "step": 6448
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.203884947846701e-06,
+      "loss": 0.076,
+      "step": 6449
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.199451116052193e-06,
+      "loss": 0.0551,
+      "step": 6450
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.195018032043026e-06,
+      "loss": 0.1834,
+      "step": 6451
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.190585696306666e-06,
+      "loss": 0.1198,
+      "step": 6452
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.186154109330485e-06,
+      "loss": 0.0953,
+      "step": 6453
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.181723271601771e-06,
+      "loss": 0.0872,
+      "step": 6454
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.177293183607747e-06,
+      "loss": 0.0996,
+      "step": 6455
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.172863845835537e-06,
+      "loss": 0.106,
+      "step": 6456
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.168435258772201e-06,
+      "loss": 0.1112,
+      "step": 6457
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.164007422904695e-06,
+      "loss": 0.0693,
+      "step": 6458
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.159580338719896e-06,
+      "loss": 0.1386,
+      "step": 6459
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.15515400670461e-06,
+      "loss": 0.1024,
+      "step": 6460
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.150728427345564e-06,
+      "loss": 0.1173,
+      "step": 6461
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.14630360112938e-06,
+      "loss": 0.0759,
+      "step": 6462
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.141879528542616e-06,
+      "loss": 0.0857,
+      "step": 6463
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.137456210071736e-06,
+      "loss": 0.0862,
+      "step": 6464
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.133033646203136e-06,
+      "loss": 0.0825,
+      "step": 6465
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.128611837423105e-06,
+      "loss": 0.0706,
+      "step": 6466
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.124190784217872e-06,
+      "loss": 0.1162,
+      "step": 6467
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.119770487073578e-06,
+      "loss": 0.0925,
+      "step": 6468
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.115350946476272e-06,
+      "loss": 0.1245,
+      "step": 6469
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.11093216291192e-06,
+      "loss": 0.1207,
+      "step": 6470
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.10651413686641e-06,
+      "loss": 0.0882,
+      "step": 6471
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.102096868825556e-06,
+      "loss": 0.1023,
+      "step": 6472
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.097680359275075e-06,
+      "loss": 0.0784,
+      "step": 6473
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.093264608700596e-06,
+      "loss": 0.1239,
+      "step": 6474
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.088849617587685e-06,
+      "loss": 0.0778,
+      "step": 6475
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.084435386421805e-06,
+      "loss": 0.0762,
+      "step": 6476
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.080021915688347e-06,
+      "loss": 0.1165,
+      "step": 6477
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.075609205872609e-06,
+      "loss": 0.0887,
+      "step": 6478
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.07119725745981e-06,
+      "loss": 0.1355,
+      "step": 6479
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.066786070935098e-06,
+      "loss": 0.0723,
+      "step": 6480
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.06237564678351e-06,
+      "loss": 0.0677,
+      "step": 6481
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.057965985490028e-06,
+      "loss": 0.0897,
+      "step": 6482
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.053557087539521e-06,
+      "loss": 0.1207,
+      "step": 6483
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.0491489534168e-06,
+      "loss": 0.1649,
+      "step": 6484
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.044741583606583e-06,
+      "loss": 0.0975,
+      "step": 6485
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.040334978593501e-06,
+      "loss": 0.0648,
+      "step": 6486
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.035929138862093e-06,
+      "loss": 0.0996,
+      "step": 6487
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.03152406489683e-06,
+      "loss": 0.1499,
+      "step": 6488
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.027119757182093e-06,
+      "loss": 0.0922,
+      "step": 6489
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.022716216202184e-06,
+      "loss": 0.0585,
+      "step": 6490
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.018313442441305e-06,
+      "loss": 0.0834,
+      "step": 6491
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.013911436383581e-06,
+      "loss": 0.1098,
+      "step": 6492
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.009510198513058e-06,
+      "loss": 0.101,
+      "step": 6493
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.005109729313703e-06,
+      "loss": 0.1275,
+      "step": 6494
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.000710029269374e-06,
+      "loss": 0.0644,
+      "step": 6495
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.996311098863875e-06,
+      "loss": 0.0745,
+      "step": 6496
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.991912938580897e-06,
+      "loss": 0.1314,
+      "step": 6497
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.98751554890407e-06,
+      "loss": 0.1347,
+      "step": 6498
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.983118930316924e-06,
+      "loss": 0.1023,
+      "step": 6499
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.978723083302909e-06,
+      "loss": 0.1199,
+      "step": 6500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.974328008345388e-06,
+      "loss": 0.086,
+      "step": 6501
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.969933705927653e-06,
+      "loss": 0.1237,
+      "step": 6502
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.965540176532892e-06,
+      "loss": 0.1346,
+      "step": 6503
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.961147420644211e-06,
+      "loss": 0.0874,
+      "step": 6504
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.95675543874464e-06,
+      "loss": 0.0738,
+      "step": 6505
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.952364231317122e-06,
+      "loss": 0.1335,
+      "step": 6506
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.947973798844506e-06,
+      "loss": 0.0931,
+      "step": 6507
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.94358414180957e-06,
+      "loss": 0.1554,
+      "step": 6508
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.939195260694987e-06,
+      "loss": 0.1138,
+      "step": 6509
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.934807155983367e-06,
+      "loss": 0.16,
+      "step": 6510
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.930419828157224e-06,
+      "loss": 0.1299,
+      "step": 6511
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.92603327769898e-06,
+      "loss": 0.0542,
+      "step": 6512
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.921647505090985e-06,
+      "loss": 0.0706,
+      "step": 6513
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.91726251081549e-06,
+      "loss": 0.1644,
+      "step": 6514
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.912878295354673e-06,
+      "loss": 0.1094,
+      "step": 6515
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.908494859190614e-06,
+      "loss": 0.0445,
+      "step": 6516
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.904112202805314e-06,
+      "loss": 0.1329,
+      "step": 6517
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.8997303266807e-06,
+      "loss": 0.112,
+      "step": 6518
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.89534923129859e-06,
+      "loss": 0.0715,
+      "step": 6519
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.890968917140724e-06,
+      "loss": 0.0856,
+      "step": 6520
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.886589384688766e-06,
+      "loss": 0.0993,
+      "step": 6521
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.882210634424287e-06,
+      "loss": 0.0782,
+      "step": 6522
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.877832666828777e-06,
+      "loss": 0.0775,
+      "step": 6523
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.873455482383629e-06,
+      "loss": 0.165,
+      "step": 6524
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.869079081570152e-06,
+      "loss": 0.0719,
+      "step": 6525
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.864703464869579e-06,
+      "loss": 0.0685,
+      "step": 6526
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.860328632763058e-06,
+      "loss": 0.1466,
+      "step": 6527
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.855954585731628e-06,
+      "loss": 0.0607,
+      "step": 6528
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.851581324256268e-06,
+      "loss": 0.0754,
+      "step": 6529
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.84720884881786e-06,
+      "loss": 0.1463,
+      "step": 6530
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.842837159897196e-06,
+      "loss": 0.1105,
+      "step": 6531
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.838466257974984e-06,
+      "loss": 0.1695,
+      "step": 6532
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.834096143531846e-06,
+      "loss": 0.1534,
+      "step": 6533
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.82972681704832e-06,
+      "loss": 0.0776,
+      "step": 6534
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.825358279004862e-06,
+      "loss": 0.104,
+      "step": 6535
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.820990529881826e-06,
+      "loss": 0.1075,
+      "step": 6536
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.816623570159483e-06,
+      "loss": 0.0792,
+      "step": 6537
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.812257400318029e-06,
+      "loss": 0.0409,
+      "step": 6538
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.807892020837571e-06,
+      "loss": 0.1275,
+      "step": 6539
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.803527432198115e-06,
+      "loss": 0.0798,
+      "step": 6540
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.799163634879597e-06,
+      "loss": 0.0347,
+      "step": 6541
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.794800629361848e-06,
+      "loss": 0.1076,
+      "step": 6542
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.790438416124629e-06,
+      "loss": 0.1461,
+      "step": 6543
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.786076995647613e-06,
+      "loss": 0.1123,
+      "step": 6544
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.781716368410368e-06,
+      "loss": 0.1258,
+      "step": 6545
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.777356534892394e-06,
+      "loss": 0.0696,
+      "step": 6546
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.77299749557309e-06,
+      "loss": 0.0829,
+      "step": 6547
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.768639250931787e-06,
+      "loss": 0.1538,
+      "step": 6548
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.764281801447698e-06,
+      "loss": 0.125,
+      "step": 6549
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.759925147599974e-06,
+      "loss": 0.084,
+      "step": 6550
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.755569289867674e-06,
+      "loss": 0.0776,
+      "step": 6551
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.751214228729768e-06,
+      "loss": 0.1111,
+      "step": 6552
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.746859964665136e-06,
+      "loss": 0.0681,
+      "step": 6553
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.742506498152557e-06,
+      "loss": 0.1326,
+      "step": 6554
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.73815382967075e-06,
+      "loss": 0.0966,
+      "step": 6555
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.733801959698334e-06,
+      "loss": 0.0822,
+      "step": 6556
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.729450888713829e-06,
+      "loss": 0.1343,
+      "step": 6557
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.725100617195687e-06,
+      "loss": 0.0691,
+      "step": 6558
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.720751145622252e-06,
+      "loss": 0.138,
+      "step": 6559
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.716402474471799e-06,
+      "loss": 0.2145,
+      "step": 6560
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.712054604222496e-06,
+      "loss": 0.1008,
+      "step": 6561
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.707707535352438e-06,
+      "loss": 0.0793,
+      "step": 6562
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.703361268339634e-06,
+      "loss": 0.0713,
+      "step": 6563
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.69901580366199e-06,
+      "loss": 0.0615,
+      "step": 6564
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.694671141797329e-06,
+      "loss": 0.0796,
+      "step": 6565
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.690327283223388e-06,
+      "loss": 0.0525,
+      "step": 6566
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.685984228417822e-06,
+      "loss": 0.0316,
+      "step": 6567
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.681641977858194e-06,
+      "loss": 0.0726,
+      "step": 6568
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.677300532021967e-06,
+      "loss": 0.1132,
+      "step": 6569
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.672959891386525e-06,
+      "loss": 0.0843,
+      "step": 6570
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.668620056429167e-06,
+      "loss": 0.0747,
+      "step": 6571
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.664281027627094e-06,
+      "loss": 0.0718,
+      "step": 6572
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.659942805457437e-06,
+      "loss": 0.1046,
+      "step": 6573
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.655605390397216e-06,
+      "loss": 0.037,
+      "step": 6574
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.651268782923363e-06,
+      "loss": 0.0896,
+      "step": 6575
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.646932983512739e-06,
+      "loss": 0.1201,
+      "step": 6576
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.642597992642113e-06,
+      "loss": 0.1169,
+      "step": 6577
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.63826381078814e-06,
+      "loss": 0.0826,
+      "step": 6578
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.633930438427418e-06,
+      "loss": 0.091,
+      "step": 6579
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.629597876036447e-06,
+      "loss": 0.0541,
+      "step": 6580
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.625266124091627e-06,
+      "loss": 0.1143,
+      "step": 6581
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.620935183069267e-06,
+      "loss": 0.1333,
+      "step": 6582
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.616605053445605e-06,
+      "loss": 0.1501,
+      "step": 6583
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.61227573569678e-06,
+      "loss": 0.1056,
+      "step": 6584
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.607947230298843e-06,
+      "loss": 0.0874,
+      "step": 6585
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.603619537727755e-06,
+      "loss": 0.0939,
+      "step": 6586
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.5992926584593795e-06,
+      "loss": 0.0504,
+      "step": 6587
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.594966592969501e-06,
+      "loss": 0.1215,
+      "step": 6588
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.590641341733821e-06,
+      "loss": 0.1371,
+      "step": 6589
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.5863169052279284e-06,
+      "loss": 0.0584,
+      "step": 6590
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.58199328392735e-06,
+      "loss": 0.0775,
+      "step": 6591
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.577670478307494e-06,
+      "loss": 0.0607,
+      "step": 6592
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.573348488843711e-06,
+      "loss": 0.1218,
+      "step": 6593
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.5690273160112285e-06,
+      "loss": 0.0901,
+      "step": 6594
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.564706960285209e-06,
+      "loss": 0.0976,
+      "step": 6595
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.5603874221407235e-06,
+      "loss": 0.0921,
+      "step": 6596
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.556068702052741e-06,
+      "loss": 0.1102,
+      "step": 6597
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.551750800496136e-06,
+      "loss": 0.1129,
+      "step": 6598
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.547433717945713e-06,
+      "loss": 0.1413,
+      "step": 6599
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.543117454876174e-06,
+      "loss": 0.0994,
+      "step": 6600
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.538802011762137e-06,
+      "loss": 0.0919,
+      "step": 6601
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.534487389078133e-06,
+      "loss": 0.1091,
+      "step": 6602
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.530173587298577e-06,
+      "loss": 0.0892,
+      "step": 6603
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.5258606068978234e-06,
+      "loss": 0.1466,
+      "step": 6604
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.521548448350123e-06,
+      "loss": 0.0547,
+      "step": 6605
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.517237112129646e-06,
+      "loss": 0.0878,
+      "step": 6606
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.512926598710454e-06,
+      "loss": 0.1045,
+      "step": 6607
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.50861690856654e-06,
+      "loss": 0.111,
+      "step": 6608
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.504308042171787e-06,
+      "loss": 0.0775,
+      "step": 6609
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.500000000000004e-06,
+      "loss": 0.1198,
+      "step": 6610
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.49569278252489e-06,
+      "loss": 0.0845,
+      "step": 6611
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.491386390220073e-06,
+      "loss": 0.1251,
+      "step": 6612
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.487080823559085e-06,
+      "loss": 0.0927,
+      "step": 6613
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.48277608301536e-06,
+      "loss": 0.1104,
+      "step": 6614
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.47847216906224e-06,
+      "loss": 0.0945,
+      "step": 6615
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.474169082172986e-06,
+      "loss": 0.0723,
+      "step": 6616
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.469866822820764e-06,
+      "loss": 0.1261,
+      "step": 6617
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.465565391478656e-06,
+      "loss": 0.0853,
+      "step": 6618
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.4612647886196366e-06,
+      "loss": 0.0919,
+      "step": 6619
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.456965014716594e-06,
+      "loss": 0.1184,
+      "step": 6620
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.452666070242335e-06,
+      "loss": 0.12,
+      "step": 6621
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.448367955669576e-06,
+      "loss": 0.1056,
+      "step": 6622
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.444070671470924e-06,
+      "loss": 0.0957,
+      "step": 6623
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.439774218118917e-06,
+      "loss": 0.0968,
+      "step": 6624
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.4354785960859815e-06,
+      "loss": 0.0705,
+      "step": 6625
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.431183805844471e-06,
+      "loss": 0.1055,
+      "step": 6626
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.42688984786663e-06,
+      "loss": 0.1067,
+      "step": 6627
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.422596722624623e-06,
+      "loss": 0.0828,
+      "step": 6628
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.418304430590524e-06,
+      "loss": 0.084,
+      "step": 6629
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.414012972236315e-06,
+      "loss": 0.1143,
+      "step": 6630
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.409722348033877e-06,
+      "loss": 0.0781,
+      "step": 6631
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.405432558454999e-06,
+      "loss": 0.0906,
+      "step": 6632
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.40114360397139e-06,
+      "loss": 0.1199,
+      "step": 6633
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.396855485054665e-06,
+      "loss": 0.1483,
+      "step": 6634
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.392568202176344e-06,
+      "loss": 0.0792,
+      "step": 6635
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.388281755807853e-06,
+      "loss": 0.1274,
+      "step": 6636
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.383996146420521e-06,
+      "loss": 0.0482,
+      "step": 6637
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.379711374485596e-06,
+      "loss": 0.146,
+      "step": 6638
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.375427440474238e-06,
+      "loss": 0.0636,
+      "step": 6639
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.3711443448574925e-06,
+      "loss": 0.1064,
+      "step": 6640
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.36686208810634e-06,
+      "loss": 0.0937,
+      "step": 6641
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.362580670691642e-06,
+      "loss": 0.0928,
+      "step": 6642
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.358300093084195e-06,
+      "loss": 0.1164,
+      "step": 6643
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.354020355754676e-06,
+      "loss": 0.132,
+      "step": 6644
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.3497414591736924e-06,
+      "loss": 0.0943,
+      "step": 6645
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.34546340381175e-06,
+      "loss": 0.0995,
+      "step": 6646
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.341186190139261e-06,
+      "loss": 0.135,
+      "step": 6647
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.336909818626537e-06,
+      "loss": 0.0654,
+      "step": 6648
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.332634289743814e-06,
+      "loss": 0.0413,
+      "step": 6649
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.328359603961226e-06,
+      "loss": 0.0909,
+      "step": 6650
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.3240857617488206e-06,
+      "loss": 0.1504,
+      "step": 6651
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.319812763576543e-06,
+      "loss": 0.1038,
+      "step": 6652
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.315540609914243e-06,
+      "loss": 0.1114,
+      "step": 6653
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.311269301231694e-06,
+      "loss": 0.0631,
+      "step": 6654
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.30699883799857e-06,
+      "loss": 0.0884,
+      "step": 6655
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.3027292206844374e-06,
+      "loss": 0.1581,
+      "step": 6656
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.298460449758789e-06,
+      "loss": 0.1284,
+      "step": 6657
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.294192525691021e-06,
+      "loss": 0.0763,
+      "step": 6658
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.289925448950426e-06,
+      "loss": 0.097,
+      "step": 6659
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.285659220006208e-06,
+      "loss": 0.0948,
+      "step": 6660
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.281393839327481e-06,
+      "loss": 0.0793,
+      "step": 6661
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.277129307383269e-06,
+      "loss": 0.151,
+      "step": 6662
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.272865624642499e-06,
+      "loss": 0.1086,
+      "step": 6663
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.268602791574002e-06,
+      "loss": 0.1014,
+      "step": 6664
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.264340808646507e-06,
+      "loss": 0.132,
+      "step": 6665
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.26007967632867e-06,
+      "loss": 0.0779,
+      "step": 6666
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.25581939508904e-06,
+      "loss": 0.1142,
+      "step": 6667
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.251559965396083e-06,
+      "loss": 0.0472,
+      "step": 6668
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.247301387718159e-06,
+      "loss": 0.0801,
+      "step": 6669
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.243043662523532e-06,
+      "loss": 0.0677,
+      "step": 6670
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.238786790280385e-06,
+      "loss": 0.1053,
+      "step": 6671
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.234530771456809e-06,
+      "loss": 0.1176,
+      "step": 6672
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.230275606520782e-06,
+      "loss": 0.1208,
+      "step": 6673
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.22602129594021e-06,
+      "loss": 0.0981,
+      "step": 6674
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.2217678401828865e-06,
+      "loss": 0.1038,
+      "step": 6675
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.217515239716529e-06,
+      "loss": 0.0848,
+      "step": 6676
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.213263495008741e-06,
+      "loss": 0.0525,
+      "step": 6677
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.209012606527046e-06,
+      "loss": 0.073,
+      "step": 6678
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.204762574738873e-06,
+      "loss": 0.0937,
+      "step": 6679
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.200513400111564e-06,
+      "loss": 0.0967,
+      "step": 6680
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.196265083112333e-06,
+      "loss": 0.1294,
+      "step": 6681
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.192017624208335e-06,
+      "loss": 0.0949,
+      "step": 6682
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.187771023866617e-06,
+      "loss": 0.0948,
+      "step": 6683
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.183525282554142e-06,
+      "loss": 0.0734,
+      "step": 6684
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.1792804007377565e-06,
+      "loss": 0.0753,
+      "step": 6685
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.175036378884236e-06,
+      "loss": 0.1527,
+      "step": 6686
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.170793217460242e-06,
+      "loss": 0.1272,
+      "step": 6687
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.166550916932363e-06,
+      "loss": 0.0673,
+      "step": 6688
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.162309477767066e-06,
+      "loss": 0.0808,
+      "step": 6689
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.1580689004307454e-06,
+      "loss": 0.1345,
+      "step": 6690
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.153829185389697e-06,
+      "loss": 0.0813,
+      "step": 6691
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.149590333110109e-06,
+      "loss": 0.1027,
+      "step": 6692
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.145352344058096e-06,
+      "loss": 0.0702,
+      "step": 6693
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.141115218699651e-06,
+      "loss": 0.131,
+      "step": 6694
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.136878957500693e-06,
+      "loss": 0.113,
+      "step": 6695
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.132643560927045e-06,
+      "loss": 0.08,
+      "step": 6696
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.1284090294444245e-06,
+      "loss": 0.1293,
+      "step": 6697
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.124175363518453e-06,
+      "loss": 0.0911,
+      "step": 6698
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.119942563614666e-06,
+      "loss": 0.1066,
+      "step": 6699
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.115710630198504e-06,
+      "loss": 0.1189,
+      "step": 6700
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.111479563735313e-06,
+      "loss": 0.1113,
+      "step": 6701
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.10724936469033e-06,
+      "loss": 0.1074,
+      "step": 6702
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.103020033528705e-06,
+      "loss": 0.1073,
+      "step": 6703
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.098791570715496e-06,
+      "loss": 0.1101,
+      "step": 6704
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.09456397671567e-06,
+      "loss": 0.0983,
+      "step": 6705
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.09033725199408e-06,
+      "loss": 0.065,
+      "step": 6706
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.086111397015499e-06,
+      "loss": 0.13,
+      "step": 6707
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.081886412244608e-06,
+      "loss": 0.1028,
+      "step": 6708
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.077662298145976e-06,
+      "loss": 0.069,
+      "step": 6709
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.073439055184081e-06,
+      "loss": 0.1031,
+      "step": 6710
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.069216683823315e-06,
+      "loss": 0.1036,
+      "step": 6711
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.064995184527969e-06,
+      "loss": 0.0949,
+      "step": 6712
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.060774557762239e-06,
+      "loss": 0.0812,
+      "step": 6713
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.05655480399022e-06,
+      "loss": 0.0521,
+      "step": 6714
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.052335923675909e-06,
+      "loss": 0.0996,
+      "step": 6715
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.0481179172832165e-06,
+      "loss": 0.1056,
+      "step": 6716
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.043900785275959e-06,
+      "loss": 0.0583,
+      "step": 6717
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.039684528117839e-06,
+      "loss": 0.0353,
+      "step": 6718
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.035469146272487e-06,
+      "loss": 0.082,
+      "step": 6719
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.0312546402034104e-06,
+      "loss": 0.0842,
+      "step": 6720
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.027041010374042e-06,
+      "loss": 0.0869,
+      "step": 6721
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.022828257247717e-06,
+      "loss": 0.103,
+      "step": 6722
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.018616381287655e-06,
+      "loss": 0.0886,
+      "step": 6723
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.014405382957003e-06,
+      "loss": 0.1404,
+      "step": 6724
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.0101952627187915e-06,
+      "loss": 0.092,
+      "step": 6725
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.005986021035971e-06,
+      "loss": 0.0948,
+      "step": 6726
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.001777658371381e-06,
+      "loss": 0.0875,
+      "step": 6727
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.997570175187775e-06,
+      "loss": 0.1252,
+      "step": 6728
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.993363571947804e-06,
+      "loss": 0.1021,
+      "step": 6729
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.9891578491140376e-06,
+      "loss": 0.1013,
+      "step": 6730
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.984953007148913e-06,
+      "loss": 0.1304,
+      "step": 6731
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.980749046514801e-06,
+      "loss": 0.1253,
+      "step": 6732
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.976545967673972e-06,
+      "loss": 0.0744,
+      "step": 6733
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.972343771088598e-06,
+      "loss": 0.077,
+      "step": 6734
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.968142457220739e-06,
+      "loss": 0.0611,
+      "step": 6735
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.96394202653238e-06,
+      "loss": 0.0713,
+      "step": 6736
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.959742479485389e-06,
+      "loss": 0.1156,
+      "step": 6737
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.955543816541555e-06,
+      "loss": 0.1272,
+      "step": 6738
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.951346038162555e-06,
+      "loss": 0.0883,
+      "step": 6739
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.947149144809975e-06,
+      "loss": 0.1207,
+      "step": 6740
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.942953136945313e-06,
+      "loss": 0.097,
+      "step": 6741
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.938758015029952e-06,
+      "loss": 0.0668,
+      "step": 6742
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.934563779525182e-06,
+      "loss": 0.1109,
+      "step": 6743
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.930370430892203e-06,
+      "loss": 0.0585,
+      "step": 6744
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.926177969592115e-06,
+      "loss": 0.0881,
+      "step": 6745
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.921986396085926e-06,
+      "loss": 0.1079,
+      "step": 6746
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.91779571083453e-06,
+      "loss": 0.0928,
+      "step": 6747
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.913605914298731e-06,
+      "loss": 0.149,
+      "step": 6748
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.909417006939241e-06,
+      "loss": 0.1085,
+      "step": 6749
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.905228989216677e-06,
+      "loss": 0.1252,
+      "step": 6750
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.901041861591541e-06,
+      "loss": 0.1072,
+      "step": 6751
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.896855624524257e-06,
+      "loss": 0.1237,
+      "step": 6752
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.892670278475132e-06,
+      "loss": 0.0932,
+      "step": 6753
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.888485823904388e-06,
+      "loss": 0.0731,
+      "step": 6754
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.884302261272154e-06,
+      "loss": 0.0828,
+      "step": 6755
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.880119591038443e-06,
+      "loss": 0.0718,
+      "step": 6756
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.8759378136631806e-06,
+      "loss": 0.1647,
+      "step": 6757
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.871756929606202e-06,
+      "loss": 0.1049,
+      "step": 6758
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.867576939327228e-06,
+      "loss": 0.0966,
+      "step": 6759
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.863397843285884e-06,
+      "loss": 0.0751,
+      "step": 6760
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.859219641941705e-06,
+      "loss": 0.0985,
+      "step": 6761
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.855042335754129e-06,
+      "loss": 0.1458,
+      "step": 6762
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.8508659251824914e-06,
+      "loss": 0.0876,
+      "step": 6763
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.846690410686025e-06,
+      "loss": 0.0916,
+      "step": 6764
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.842515792723863e-06,
+      "loss": 0.0608,
+      "step": 6765
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.838342071755049e-06,
+      "loss": 0.0463,
+      "step": 6766
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.834169248238526e-06,
+      "loss": 0.0845,
+      "step": 6767
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.829997322633129e-06,
+      "loss": 0.1683,
+      "step": 6768
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.825826295397613e-06,
+      "loss": 0.1015,
+      "step": 6769
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.821656166990606e-06,
+      "loss": 0.1165,
+      "step": 6770
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.817486937870669e-06,
+      "loss": 0.1414,
+      "step": 6771
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.813318608496237e-06,
+      "loss": 0.0756,
+      "step": 6772
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.809151179325661e-06,
+      "loss": 0.1096,
+      "step": 6773
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.8049846508171954e-06,
+      "loss": 0.1104,
+      "step": 6774
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.800819023428988e-06,
+      "loss": 0.0543,
+      "step": 6775
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.79665429761908e-06,
+      "loss": 0.1287,
+      "step": 6776
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.792490473845428e-06,
+      "loss": 0.0939,
+      "step": 6777
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.788327552565888e-06,
+      "loss": 0.1317,
+      "step": 6778
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.784165534238214e-06,
+      "loss": 0.1618,
+      "step": 6779
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.780004419320059e-06,
+      "loss": 0.1241,
+      "step": 6780
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.7758442082689675e-06,
+      "loss": 0.0828,
+      "step": 6781
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.771684901542403e-06,
+      "loss": 0.097,
+      "step": 6782
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.767526499597719e-06,
+      "loss": 0.1126,
+      "step": 6783
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.763369002892177e-06,
+      "loss": 0.1136,
+      "step": 6784
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.759212411882925e-06,
+      "loss": 0.108,
+      "step": 6785
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.7550567270270265e-06,
+      "loss": 0.1088,
+      "step": 6786
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.7509019487814335e-06,
+      "loss": 0.1014,
+      "step": 6787
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.74674807760301e-06,
+      "loss": 0.1194,
+      "step": 6788
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.742595113948504e-06,
+      "loss": 0.0703,
+      "step": 6789
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.738443058274579e-06,
+      "loss": 0.053,
+      "step": 6790
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.7342919110378e-06,
+      "loss": 0.1316,
+      "step": 6791
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.730141672694619e-06,
+      "loss": 0.0978,
+      "step": 6792
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.7259923437013885e-06,
+      "loss": 0.0761,
+      "step": 6793
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.721843924514374e-06,
+      "loss": 0.1319,
+      "step": 6794
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.717696415589731e-06,
+      "loss": 0.1379,
+      "step": 6795
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.713549817383524e-06,
+      "loss": 0.1075,
+      "step": 6796
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.709404130351708e-06,
+      "loss": 0.1418,
+      "step": 6797
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.705259354950133e-06,
+      "loss": 0.0812,
+      "step": 6798
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.701115491634562e-06,
+      "loss": 0.102,
+      "step": 6799
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.6969725408606595e-06,
+      "loss": 0.0959,
+      "step": 6800
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.692830503083971e-06,
+      "loss": 0.1021,
+      "step": 6801
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.688689378759965e-06,
+      "loss": 0.158,
+      "step": 6802
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.684549168343985e-06,
+      "loss": 0.0798,
+      "step": 6803
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.680409872291298e-06,
+      "loss": 0.2161,
+      "step": 6804
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.676271491057047e-06,
+      "loss": 0.0827,
+      "step": 6805
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.672134025096296e-06,
+      "loss": 0.1168,
+      "step": 6806
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.667997474863995e-06,
+      "loss": 0.0754,
+      "step": 6807
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.6638618408150095e-06,
+      "loss": 0.0604,
+      "step": 6808
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.659727123404072e-06,
+      "loss": 0.2017,
+      "step": 6809
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.6555933230858424e-06,
+      "loss": 0.0625,
+      "step": 6810
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.651460440314873e-06,
+      "loss": 0.0823,
+      "step": 6811
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.647328475545611e-06,
+      "loss": 0.0612,
+      "step": 6812
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.643197429232416e-06,
+      "loss": 0.1336,
+      "step": 6813
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.639067301829526e-06,
+      "loss": 0.0625,
+      "step": 6814
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.6349380937910845e-06,
+      "loss": 0.1378,
+      "step": 6815
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.630809805571145e-06,
+      "loss": 0.1113,
+      "step": 6816
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.626682437623653e-06,
+      "loss": 0.0815,
+      "step": 6817
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.622555990402443e-06,
+      "loss": 0.0936,
+      "step": 6818
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.61843046436127e-06,
+      "loss": 0.0823,
+      "step": 6819
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.614305859953762e-06,
+      "loss": 0.1029,
+      "step": 6820
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.61018217763347e-06,
+      "loss": 0.1187,
+      "step": 6821
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.6060594178538226e-06,
+      "loss": 0.0865,
+      "step": 6822
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.601937581068161e-06,
+      "loss": 0.0762,
+      "step": 6823
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.597816667729727e-06,
+      "loss": 0.0823,
+      "step": 6824
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.593696678291646e-06,
+      "loss": 0.1033,
+      "step": 6825
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.58957761320695e-06,
+      "loss": 0.0665,
+      "step": 6826
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.585459472928569e-06,
+      "loss": 0.0524,
+      "step": 6827
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.581342257909337e-06,
+      "loss": 0.0996,
+      "step": 6828
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.577225968601983e-06,
+      "loss": 0.0859,
+      "step": 6829
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.573110605459129e-06,
+      "loss": 0.0769,
+      "step": 6830
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.568996168933292e-06,
+      "loss": 0.1319,
+      "step": 6831
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.5648826594769e-06,
+      "loss": 0.0712,
+      "step": 6832
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.560770077542278e-06,
+      "loss": 0.0381,
+      "step": 6833
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.55665842358163e-06,
+      "loss": 0.1233,
+      "step": 6834
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.552547698047082e-06,
+      "loss": 0.0893,
+      "step": 6835
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.548437901390649e-06,
+      "loss": 0.089,
+      "step": 6836
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.544329034064238e-06,
+      "loss": 0.0324,
+      "step": 6837
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.5402210965196555e-06,
+      "loss": 0.1179,
+      "step": 6838
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.53611408920861e-06,
+      "loss": 0.1441,
+      "step": 6839
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.532008012582707e-06,
+      "loss": 0.0991,
+      "step": 6840
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.5279028670934575e-06,
+      "loss": 0.0936,
+      "step": 6841
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.523798653192253e-06,
+      "loss": 0.1066,
+      "step": 6842
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.519695371330385e-06,
+      "loss": 0.0873,
+      "step": 6843
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.5155930219590555e-06,
+      "loss": 0.0819,
+      "step": 6844
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.511491605529357e-06,
+      "loss": 0.0878,
+      "step": 6845
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.507391122492287e-06,
+      "loss": 0.0975,
+      "step": 6846
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.503291573298722e-06,
+      "loss": 0.0859,
+      "step": 6847
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.499192958399446e-06,
+      "loss": 0.1301,
+      "step": 6848
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.495095278245143e-06,
+      "loss": 0.0754,
+      "step": 6849
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.490998533286401e-06,
+      "loss": 0.1427,
+      "step": 6850
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.486902723973681e-06,
+      "loss": 0.0839,
+      "step": 6851
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.4828078507573714e-06,
+      "loss": 0.0622,
+      "step": 6852
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.4787139140877295e-06,
+      "loss": 0.0799,
+      "step": 6853
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.474620914414935e-06,
+      "loss": 0.1136,
+      "step": 6854
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.47052885218904e-06,
+      "loss": 0.1442,
+      "step": 6855
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.466437727860013e-06,
+      "loss": 0.0903,
+      "step": 6856
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.4623475418777145e-06,
+      "loss": 0.083,
+      "step": 6857
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.458258294691898e-06,
+      "loss": 0.0873,
+      "step": 6858
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.454169986752208e-06,
+      "loss": 0.1155,
+      "step": 6859
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.4500826185081975e-06,
+      "loss": 0.0834,
+      "step": 6860
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.445996190409313e-06,
+      "loss": 0.1719,
+      "step": 6861
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.4419107029049016e-06,
+      "loss": 0.0525,
+      "step": 6862
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.437826156444192e-06,
+      "loss": 0.0422,
+      "step": 6863
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.433742551476326e-06,
+      "loss": 0.1503,
+      "step": 6864
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.429659888450329e-06,
+      "loss": 0.1068,
+      "step": 6865
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.425578167815136e-06,
+      "loss": 0.0651,
+      "step": 6866
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.421497390019564e-06,
+      "loss": 0.146,
+      "step": 6867
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.417417555512335e-06,
+      "loss": 0.064,
+      "step": 6868
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.4133386647420734e-06,
+      "loss": 0.0749,
+      "step": 6869
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.409260718157287e-06,
+      "loss": 0.1434,
+      "step": 6870
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.405183716206379e-06,
+      "loss": 0.1327,
+      "step": 6871
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.401107659337659e-06,
+      "loss": 0.0839,
+      "step": 6872
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.397032547999331e-06,
+      "loss": 0.0448,
+      "step": 6873
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.392958382639494e-06,
+      "loss": 0.1617,
+      "step": 6874
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.388885163706142e-06,
+      "loss": 0.1006,
+      "step": 6875
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.384812891647153e-06,
+      "loss": 0.076,
+      "step": 6876
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.380741566910321e-06,
+      "loss": 0.0904,
+      "step": 6877
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.3766711899433235e-06,
+      "loss": 0.0682,
+      "step": 6878
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.3726017611937475e-06,
+      "loss": 0.1186,
+      "step": 6879
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.368533281109059e-06,
+      "loss": 0.1189,
+      "step": 6880
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.364465750136618e-06,
+      "loss": 0.0723,
+      "step": 6881
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.360399168723695e-06,
+      "loss": 0.0966,
+      "step": 6882
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.356333537317458e-06,
+      "loss": 0.1183,
+      "step": 6883
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.3522688563649465e-06,
+      "loss": 0.0774,
+      "step": 6884
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.348205126313119e-06,
+      "loss": 0.1019,
+      "step": 6885
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.344142347608828e-06,
+      "loss": 0.0943,
+      "step": 6886
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.3400805206988066e-06,
+      "loss": 0.1122,
+      "step": 6887
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.336019646029688e-06,
+      "loss": 0.1292,
+      "step": 6888
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.331959724048009e-06,
+      "loss": 0.0778,
+      "step": 6889
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.327900755200197e-06,
+      "loss": 0.0641,
+      "step": 6890
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.3238427399325795e-06,
+      "loss": 0.0443,
+      "step": 6891
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.31978567869137e-06,
+      "loss": 0.0598,
+      "step": 6892
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.315729571922672e-06,
+      "loss": 0.1244,
+      "step": 6893
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.311674420072503e-06,
+      "loss": 0.0773,
+      "step": 6894
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.307620223586769e-06,
+      "loss": 0.1012,
+      "step": 6895
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.303566982911258e-06,
+      "loss": 0.1199,
+      "step": 6896
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.2995146984916714e-06,
+      "loss": 0.103,
+      "step": 6897
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.295463370773586e-06,
+      "loss": 0.0959,
+      "step": 6898
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.291413000202497e-06,
+      "loss": 0.1336,
+      "step": 6899
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.287363587223769e-06,
+      "loss": 0.1004,
+      "step": 6900
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.28331513228268e-06,
+      "loss": 0.0785,
+      "step": 6901
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.279267635824401e-06,
+      "loss": 0.0678,
+      "step": 6902
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.27522109829398e-06,
+      "loss": 0.0629,
+      "step": 6903
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.271175520136387e-06,
+      "loss": 0.079,
+      "step": 6904
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.26713090179646e-06,
+      "loss": 0.0776,
+      "step": 6905
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.263087243718947e-06,
+      "loss": 0.0815,
+      "step": 6906
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.259044546348495e-06,
+      "loss": 0.0805,
+      "step": 6907
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.25500281012963e-06,
+      "loss": 0.1241,
+      "step": 6908
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.250962035506775e-06,
+      "loss": 0.153,
+      "step": 6909
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.246922222924257e-06,
+      "loss": 0.1027,
+      "step": 6910
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.24288337282629e-06,
+      "loss": 0.0849,
+      "step": 6911
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.238845485656992e-06,
+      "loss": 0.075,
+      "step": 6912
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.234808561860358e-06,
+      "loss": 0.1081,
+      "step": 6913
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.230772601880292e-06,
+      "loss": 0.1155,
+      "step": 6914
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.226737606160581e-06,
+      "loss": 0.1073,
+      "step": 6915
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.222703575144918e-06,
+      "loss": 0.0646,
+      "step": 6916
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.218670509276874e-06,
+      "loss": 0.1371,
+      "step": 6917
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.214638408999929e-06,
+      "loss": 0.0933,
+      "step": 6918
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.210607274757458e-06,
+      "loss": 0.1053,
+      "step": 6919
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.206577106992714e-06,
+      "loss": 0.0999,
+      "step": 6920
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.202547906148849e-06,
+      "loss": 0.1471,
+      "step": 6921
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.198519672668916e-06,
+      "loss": 0.088,
+      "step": 6922
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.194492406995859e-06,
+      "loss": 0.0943,
+      "step": 6923
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.19046610957252e-06,
+      "loss": 0.0623,
+      "step": 6924
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.186440780841623e-06,
+      "loss": 0.135,
+      "step": 6925
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.1824164212457866e-06,
+      "loss": 0.0921,
+      "step": 6926
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.178393031227531e-06,
+      "loss": 0.1076,
+      "step": 6927
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.174370611229274e-06,
+      "loss": 0.0645,
+      "step": 6928
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.170349161693307e-06,
+      "loss": 0.0846,
+      "step": 6929
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.1663286830618365e-06,
+      "loss": 0.1009,
+      "step": 6930
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.162309175776946e-06,
+      "loss": 0.1135,
+      "step": 6931
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.15829064028062e-06,
+      "loss": 0.0777,
+      "step": 6932
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.154273077014743e-06,
+      "loss": 0.0822,
+      "step": 6933
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.150256486421071e-06,
+      "loss": 0.0938,
+      "step": 6934
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.146240868941278e-06,
+      "loss": 0.076,
+      "step": 6935
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.142226225016911e-06,
+      "loss": 0.141,
+      "step": 6936
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.138212555089428e-06,
+      "loss": 0.0746,
+      "step": 6937
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.1341998596001585e-06,
+      "loss": 0.0821,
+      "step": 6938
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.1301881389903434e-06,
+      "loss": 0.1409,
+      "step": 6939
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.126177393701108e-06,
+      "loss": 0.1291,
+      "step": 6940
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.122167624173481e-06,
+      "loss": 0.0609,
+      "step": 6941
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.118158830848366e-06,
+      "loss": 0.139,
+      "step": 6942
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.114151014166563e-06,
+      "loss": 0.1141,
+      "step": 6943
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.110144174568777e-06,
+      "loss": 0.0759,
+      "step": 6944
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.106138312495602e-06,
+      "loss": 0.1199,
+      "step": 6945
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.102133428387511e-06,
+      "loss": 0.087,
+      "step": 6946
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.098129522684891e-06,
+      "loss": 0.0782,
+      "step": 6947
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.094126595827997e-06,
+      "loss": 0.076,
+      "step": 6948
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.090124648257e-06,
+      "loss": 0.1522,
+      "step": 6949
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.086123680411943e-06,
+      "loss": 0.0986,
+      "step": 6950
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.082123692732776e-06,
+      "loss": 0.0775,
+      "step": 6951
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.078124685659339e-06,
+      "loss": 0.0861,
+      "step": 6952
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.0741266596313574e-06,
+      "loss": 0.0971,
+      "step": 6953
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.070129615088448e-06,
+      "loss": 0.1111,
+      "step": 6954
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.066133552470128e-06,
+      "loss": 0.0477,
+      "step": 6955
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.062138472215801e-06,
+      "loss": 0.0734,
+      "step": 6956
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.058144374764772e-06,
+      "loss": 0.101,
+      "step": 6957
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.054151260556226e-06,
+      "loss": 0.129,
+      "step": 6958
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.050159130029235e-06,
+      "loss": 0.0615,
+      "step": 6959
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.04616798362278e-06,
+      "loss": 0.1099,
+      "step": 6960
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.042177821775732e-06,
+      "loss": 0.1099,
+      "step": 6961
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.038188644926834e-06,
+      "loss": 0.1014,
+      "step": 6962
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.034200453514741e-06,
+      "loss": 0.1303,
+      "step": 6963
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.030213247977995e-06,
+      "loss": 0.0603,
+      "step": 6964
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.026227028755022e-06,
+      "loss": 0.1177,
+      "step": 6965
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.022241796284152e-06,
+      "loss": 0.1316,
+      "step": 6966
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.018257551003589e-06,
+      "loss": 0.0951,
+      "step": 6967
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.014274293351446e-06,
+      "loss": 0.0686,
+      "step": 6968
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.010292023765723e-06,
+      "loss": 0.0775,
+      "step": 6969
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.006310742684306e-06,
+      "loss": 0.0339,
+      "step": 6970
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.002330450544966e-06,
+      "loss": 0.0864,
+      "step": 6971
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.998351147785383e-06,
+      "loss": 0.0797,
+      "step": 6972
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.99437283484312e-06,
+      "loss": 0.0911,
+      "step": 6973
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.990395512155631e-06,
+      "loss": 0.049,
+      "step": 6974
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.98641918016026e-06,
+      "loss": 0.0578,
+      "step": 6975
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.982443839294235e-06,
+      "loss": 0.0581,
+      "step": 6976
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.978469489994687e-06,
+      "loss": 0.1365,
+      "step": 6977
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.974496132698642e-06,
+      "loss": 0.0533,
+      "step": 6978
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.970523767842998e-06,
+      "loss": 0.0701,
+      "step": 6979
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.966552395864563e-06,
+      "loss": 0.151,
+      "step": 6980
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.9625820172000164e-06,
+      "loss": 0.1405,
+      "step": 6981
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.958612632285952e-06,
+      "loss": 0.1078,
+      "step": 6982
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.954644241558828e-06,
+      "loss": 0.094,
+      "step": 6983
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.950676845455016e-06,
+      "loss": 0.0674,
+      "step": 6984
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.946710444410771e-06,
+      "loss": 0.071,
+      "step": 6985
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.942745038862235e-06,
+      "loss": 0.0598,
+      "step": 6986
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.938780629245433e-06,
+      "loss": 0.0755,
+      "step": 6987
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.934817215996297e-06,
+      "loss": 0.0956,
+      "step": 6988
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.930854799550642e-06,
+      "loss": 0.0981,
+      "step": 6989
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.926893380344179e-06,
+      "loss": 0.0732,
+      "step": 6990
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.922932958812494e-06,
+      "loss": 0.0954,
+      "step": 6991
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.918973535391083e-06,
+      "loss": 0.1036,
+      "step": 6992
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.91501511051531e-06,
+      "loss": 0.1048,
+      "step": 6993
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.91105768462045e-06,
+      "loss": 0.0945,
+      "step": 6994
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.907101258141665e-06,
+      "loss": 0.0841,
+      "step": 6995
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.903145831513989e-06,
+      "loss": 0.1184,
+      "step": 6996
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.8991914051723715e-06,
+      "loss": 0.0838,
+      "step": 6997
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.895237979551627e-06,
+      "loss": 0.1061,
+      "step": 6998
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.891285555086484e-06,
+      "loss": 0.0753,
+      "step": 6999
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.887334132211541e-06,
+      "loss": 0.0764,
+      "step": 7000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.8833837113612945e-06,
+      "loss": 0.1087,
+      "step": 7001
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.879434292970142e-06,
+      "loss": 0.1042,
+      "step": 7002
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.8754858774723515e-06,
+      "loss": 0.0743,
+      "step": 7003
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.871538465302085e-06,
+      "loss": 0.0913,
+      "step": 7004
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.867592056893402e-06,
+      "loss": 0.101,
+      "step": 7005
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.86364665268025e-06,
+      "loss": 0.1081,
+      "step": 7006
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.8597022530964665e-06,
+      "loss": 0.1361,
+      "step": 7007
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.855758858575773e-06,
+      "loss": 0.0883,
+      "step": 7008
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.851816469551776e-06,
+      "loss": 0.0752,
+      "step": 7009
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.847875086457987e-06,
+      "loss": 0.1071,
+      "step": 7010
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.843934709727801e-06,
+      "loss": 0.0786,
+      "step": 7011
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.839995339794492e-06,
+      "loss": 0.1004,
+      "step": 7012
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.836056977091241e-06,
+      "loss": 0.064,
+      "step": 7013
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.832119622051095e-06,
+      "loss": 0.0637,
+      "step": 7014
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.82818327510702e-06,
+      "loss": 0.1281,
+      "step": 7015
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.824247936691841e-06,
+      "loss": 0.1133,
+      "step": 7016
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.8203136072382914e-06,
+      "loss": 0.1007,
+      "step": 7017
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.816380287178989e-06,
+      "loss": 0.0819,
+      "step": 7018
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.812447976946446e-06,
+      "loss": 0.0905,
+      "step": 7019
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.808516676973051e-06,
+      "loss": 0.1164,
+      "step": 7020
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.8045863876910825e-06,
+      "loss": 0.122,
+      "step": 7021
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.80065710953272e-06,
+      "loss": 0.0628,
+      "step": 7022
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.796728842930022e-06,
+      "loss": 0.1335,
+      "step": 7023
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.792801588314948e-06,
+      "loss": 0.0808,
+      "step": 7024
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.788875346119331e-06,
+      "loss": 0.1078,
+      "step": 7025
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.784950116774893e-06,
+      "loss": 0.0933,
+      "step": 7026
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.781025900713257e-06,
+      "loss": 0.1131,
+      "step": 7027
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.77710269836593e-06,
+      "loss": 0.1209,
+      "step": 7028
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.773180510164298e-06,
+      "loss": 0.0686,
+      "step": 7029
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.769259336539654e-06,
+      "loss": 0.1418,
+      "step": 7030
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.765339177923156e-06,
+      "loss": 0.1395,
+      "step": 7031
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.7614200347458736e-06,
+      "loss": 0.1436,
+      "step": 7032
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.757501907438744e-06,
+      "loss": 0.0769,
+      "step": 7033
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.753584796432608e-06,
+      "loss": 0.0828,
+      "step": 7034
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.7496687021581965e-06,
+      "loss": 0.0887,
+      "step": 7035
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.745753625046113e-06,
+      "loss": 0.1158,
+      "step": 7036
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.741839565526854e-06,
+      "loss": 0.0806,
+      "step": 7037
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.737926524030814e-06,
+      "loss": 0.0778,
+      "step": 7038
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.734014500988268e-06,
+      "loss": 0.1227,
+      "step": 7039
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.730103496829385e-06,
+      "loss": 0.0844,
+      "step": 7040
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.726193511984208e-06,
+      "loss": 0.1324,
+      "step": 7041
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.722284546882687e-06,
+      "loss": 0.0924,
+      "step": 7042
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.718376601954641e-06,
+      "loss": 0.1852,
+      "step": 7043
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.714469677629794e-06,
+      "loss": 0.0931,
+      "step": 7044
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.710563774337742e-06,
+      "loss": 0.09,
+      "step": 7045
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.706658892507979e-06,
+      "loss": 0.1107,
+      "step": 7046
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.702755032569889e-06,
+      "loss": 0.0677,
+      "step": 7047
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.698852194952737e-06,
+      "loss": 0.0776,
+      "step": 7048
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.694950380085669e-06,
+      "loss": 0.0861,
+      "step": 7049
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.691049588397733e-06,
+      "loss": 0.0581,
+      "step": 7050
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.6871498203178585e-06,
+      "loss": 0.088,
+      "step": 7051
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.6832510762748675e-06,
+      "loss": 0.0913,
+      "step": 7052
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.679353356697458e-06,
+      "loss": 0.0978,
+      "step": 7053
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.675456662014218e-06,
+      "loss": 0.11,
+      "step": 7054
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.671560992653629e-06,
+      "loss": 0.1237,
+      "step": 7055
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.66766634904406e-06,
+      "loss": 0.1327,
+      "step": 7056
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.663772731613769e-06,
+      "loss": 0.0581,
+      "step": 7057
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.659880140790888e-06,
+      "loss": 0.059,
+      "step": 7058
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.655988577003445e-06,
+      "loss": 0.0957,
+      "step": 7059
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.652098040679354e-06,
+      "loss": 0.0731,
+      "step": 7060
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.648208532246425e-06,
+      "loss": 0.0803,
+      "step": 7061
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.644320052132336e-06,
+      "loss": 0.148,
+      "step": 7062
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.6404326007646725e-06,
+      "loss": 0.0446,
+      "step": 7063
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.636546178570886e-06,
+      "loss": 0.0715,
+      "step": 7064
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.632660785978337e-06,
+      "loss": 0.1109,
+      "step": 7065
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.6287764234142516e-06,
+      "loss": 0.1231,
+      "step": 7066
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.624893091305756e-06,
+      "loss": 0.0944,
+      "step": 7067
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.621010790079862e-06,
+      "loss": 0.0662,
+      "step": 7068
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.61712952016347e-06,
+      "loss": 0.0967,
+      "step": 7069
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.613249281983357e-06,
+      "loss": 0.1266,
+      "step": 7070
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.609370075966187e-06,
+      "loss": 0.0782,
+      "step": 7071
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.605491902538521e-06,
+      "loss": 0.1168,
+      "step": 7072
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.601614762126806e-06,
+      "loss": 0.1164,
+      "step": 7073
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.5977386551573626e-06,
+      "loss": 0.1165,
+      "step": 7074
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.593863582056413e-06,
+      "loss": 0.1425,
+      "step": 7075
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.5899895432500515e-06,
+      "loss": 0.0902,
+      "step": 7076
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.586116539164273e-06,
+      "loss": 0.0642,
+      "step": 7077
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.582244570224942e-06,
+      "loss": 0.1024,
+      "step": 7078
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.578373636857825e-06,
+      "loss": 0.102,
+      "step": 7079
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.574503739488571e-06,
+      "loss": 0.1368,
+      "step": 7080
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.5706348785427094e-06,
+      "loss": 0.097,
+      "step": 7081
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.56676705444565e-06,
+      "loss": 0.1259,
+      "step": 7082
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.562900267622707e-06,
+      "loss": 0.0618,
+      "step": 7083
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.5590345184990664e-06,
+      "loss": 0.1168,
+      "step": 7084
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.5551698074998105e-06,
+      "loss": 0.0892,
+      "step": 7085
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.551306135049897e-06,
+      "loss": 0.1152,
+      "step": 7086
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.547443501574168e-06,
+      "loss": 0.1194,
+      "step": 7087
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.543581907497362e-06,
+      "loss": 0.0473,
+      "step": 7088
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.5397213532440975e-06,
+      "loss": 0.1021,
+      "step": 7089
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.535861839238885e-06,
+      "loss": 0.0684,
+      "step": 7090
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.532003365906106e-06,
+      "loss": 0.0828,
+      "step": 7091
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.528145933670044e-06,
+      "loss": 0.0808,
+      "step": 7092
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.524289542954854e-06,
+      "loss": 0.0737,
+      "step": 7093
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.5204341941845895e-06,
+      "loss": 0.1229,
+      "step": 7094
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.5165798877831745e-06,
+      "loss": 0.0861,
+      "step": 7095
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.512726624174433e-06,
+      "loss": 0.056,
+      "step": 7096
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.508874403782068e-06,
+      "loss": 0.1375,
+      "step": 7097
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.505023227029671e-06,
+      "loss": 0.1401,
+      "step": 7098
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.5011730943407046e-06,
+      "loss": 0.0741,
+      "step": 7099
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.497324006138536e-06,
+      "loss": 0.0739,
+      "step": 7100
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.493475962846407e-06,
+      "loss": 0.1148,
+      "step": 7101
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.489628964887453e-06,
+      "loss": 0.0758,
+      "step": 7102
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.485783012684683e-06,
+      "loss": 0.1314,
+      "step": 7103
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.4819381066609906e-06,
+      "loss": 0.0774,
+      "step": 7104
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.478094247239164e-06,
+      "loss": 0.1208,
+      "step": 7105
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.4742514348418805e-06,
+      "loss": 0.0707,
+      "step": 7106
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.470409669891682e-06,
+      "loss": 0.0913,
+      "step": 7107
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.466568952811017e-06,
+      "loss": 0.1361,
+      "step": 7108
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.4627292840222e-06,
+      "loss": 0.0731,
+      "step": 7109
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.458890663947449e-06,
+      "loss": 0.1496,
+      "step": 7110
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.455053093008845e-06,
+      "loss": 0.1488,
+      "step": 7111
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.451216571628374e-06,
+      "loss": 0.0721,
+      "step": 7112
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.4473811002279e-06,
+      "loss": 0.0781,
+      "step": 7113
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.443546679229162e-06,
+      "loss": 0.0634,
+      "step": 7114
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.439713309053801e-06,
+      "loss": 0.1116,
+      "step": 7115
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.435880990123323e-06,
+      "loss": 0.1283,
+      "step": 7116
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.4320497228591315e-06,
+      "loss": 0.0532,
+      "step": 7117
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.428219507682511e-06,
+      "loss": 0.1049,
+      "step": 7118
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.424390345014635e-06,
+      "loss": 0.1029,
+      "step": 7119
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.420562235276555e-06,
+      "loss": 0.093,
+      "step": 7120
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.4167351788892e-06,
+      "loss": 0.0955,
+      "step": 7121
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.412909176273397e-06,
+      "loss": 0.0992,
+      "step": 7122
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.409084227849857e-06,
+      "loss": 0.0806,
+      "step": 7123
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.405260334039158e-06,
+      "loss": 0.0936,
+      "step": 7124
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.401437495261786e-06,
+      "loss": 0.0721,
+      "step": 7125
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.397615711938088e-06,
+      "loss": 0.1217,
+      "step": 7126
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.3937949844883166e-06,
+      "loss": 0.0728,
+      "step": 7127
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.389975313332584e-06,
+      "loss": 0.1072,
+      "step": 7128
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.386156698890906e-06,
+      "loss": 0.152,
+      "step": 7129
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.3823391415831836e-06,
+      "loss": 0.1269,
+      "step": 7130
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.378522641829186e-06,
+      "loss": 0.1254,
+      "step": 7131
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.374707200048568e-06,
+      "loss": 0.054,
+      "step": 7132
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.37089281666088e-06,
+      "loss": 0.1267,
+      "step": 7133
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.367079492085551e-06,
+      "loss": 0.1055,
+      "step": 7134
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.363267226741895e-06,
+      "loss": 0.1439,
+      "step": 7135
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.359456021049105e-06,
+      "loss": 0.0549,
+      "step": 7136
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.355645875426253e-06,
+      "loss": 0.0729,
+      "step": 7137
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.351836790292305e-06,
+      "loss": 0.0745,
+      "step": 7138
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.348028766066113e-06,
+      "loss": 0.0841,
+      "step": 7139
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.3442218031663945e-06,
+      "loss": 0.0491,
+      "step": 7140
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.340415902011773e-06,
+      "loss": 0.0937,
+      "step": 7141
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.336611063020732e-06,
+      "loss": 0.055,
+      "step": 7142
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.33280728661166e-06,
+      "loss": 0.0739,
+      "step": 7143
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.329004573202809e-06,
+      "loss": 0.0975,
+      "step": 7144
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.325202923212328e-06,
+      "loss": 0.0935,
+      "step": 7145
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.321402337058246e-06,
+      "loss": 0.1005,
+      "step": 7146
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.317602815158479e-06,
+      "loss": 0.0753,
+      "step": 7147
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.313804357930814e-06,
+      "loss": 0.1086,
+      "step": 7148
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.310006965792926e-06,
+      "loss": 0.058,
+      "step": 7149
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.306210639162374e-06,
+      "loss": 0.1032,
+      "step": 7150
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.302415378456605e-06,
+      "loss": 0.1102,
+      "step": 7151
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.2986211840929485e-06,
+      "loss": 0.0632,
+      "step": 7152
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.294828056488606e-06,
+      "loss": 0.0841,
+      "step": 7153
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.291035996060662e-06,
+      "loss": 0.0972,
+      "step": 7154
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.287245003226099e-06,
+      "loss": 0.1255,
+      "step": 7155
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.283455078401775e-06,
+      "loss": 0.0727,
+      "step": 7156
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.279666222004419e-06,
+      "loss": 0.1403,
+      "step": 7157
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.275878434450661e-06,
+      "loss": 0.1192,
+      "step": 7158
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.272091716156998e-06,
+      "loss": 0.1088,
+      "step": 7159
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.26830606753982e-06,
+      "loss": 0.0631,
+      "step": 7160
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.26452148901539e-06,
+      "loss": 0.0933,
+      "step": 7161
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.2607379809998635e-06,
+      "loss": 0.0884,
+      "step": 7162
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.2569555439092755e-06,
+      "loss": 0.073,
+      "step": 7163
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.253174178159536e-06,
+      "loss": 0.0884,
+      "step": 7164
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.2493938841664415e-06,
+      "loss": 0.0476,
+      "step": 7165
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.245614662345672e-06,
+      "loss": 0.1293,
+      "step": 7166
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.241836513112793e-06,
+      "loss": 0.054,
+      "step": 7167
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.238059436883248e-06,
+      "loss": 0.0716,
+      "step": 7168
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.234283434072357e-06,
+      "loss": 0.091,
+      "step": 7169
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.230508505095337e-06,
+      "loss": 0.1208,
+      "step": 7170
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.2267346503672654e-06,
+      "loss": 0.1604,
+      "step": 7171
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.222961870303127e-06,
+      "loss": 0.1132,
+      "step": 7172
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.219190165317761e-06,
+      "loss": 0.1174,
+      "step": 7173
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.21541953582591e-06,
+      "loss": 0.0962,
+      "step": 7174
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.211649982242195e-06,
+      "loss": 0.0638,
+      "step": 7175
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.207881504981104e-06,
+      "loss": 0.0804,
+      "step": 7176
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.204114104457028e-06,
+      "loss": 0.1005,
+      "step": 7177
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.200347781084218e-06,
+      "loss": 0.1205,
+      "step": 7178
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.1965825352768224e-06,
+      "loss": 0.1258,
+      "step": 7179
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.1928183674488714e-06,
+      "loss": 0.0792,
+      "step": 7180
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.1890552780142655e-06,
+      "loss": 0.0428,
+      "step": 7181
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.185293267386789e-06,
+      "loss": 0.1159,
+      "step": 7182
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.181532335980114e-06,
+      "loss": 0.0559,
+      "step": 7183
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.177772484207792e-06,
+      "loss": 0.1209,
+      "step": 7184
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.174013712483259e-06,
+      "loss": 0.0929,
+      "step": 7185
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.170256021219823e-06,
+      "loss": 0.0866,
+      "step": 7186
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.166499410830672e-06,
+      "loss": 0.1155,
+      "step": 7187
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.162743881728888e-06,
+      "loss": 0.1011,
+      "step": 7188
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.158989434327433e-06,
+      "loss": 0.1601,
+      "step": 7189
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.155236069039132e-06,
+      "loss": 0.074,
+      "step": 7190
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.151483786276713e-06,
+      "loss": 0.0689,
+      "step": 7191
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.147732586452769e-06,
+      "loss": 0.0823,
+      "step": 7192
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.143982469979788e-06,
+      "loss": 0.0894,
+      "step": 7193
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.140233437270119e-06,
+      "loss": 0.0752,
+      "step": 7194
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.136485488736014e-06,
+      "loss": 0.077,
+      "step": 7195
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.132738624789591e-06,
+      "loss": 0.1057,
+      "step": 7196
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.12899284584286e-06,
+      "loss": 0.1476,
+      "step": 7197
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.125248152307702e-06,
+      "loss": 0.0596,
+      "step": 7198
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.121504544595875e-06,
+      "loss": 0.1281,
+      "step": 7199
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.11776202311903e-06,
+      "loss": 0.0847,
+      "step": 7200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.114020588288697e-06,
+      "loss": 0.0959,
+      "step": 7201
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.1102802405162745e-06,
+      "loss": 0.0872,
+      "step": 7202
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.106540980213059e-06,
+      "loss": 0.0558,
+      "step": 7203
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.102802807790205e-06,
+      "loss": 0.1016,
+      "step": 7204
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.09906572365877e-06,
+      "loss": 0.0699,
+      "step": 7205
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.0953297282296836e-06,
+      "loss": 0.0749,
+      "step": 7206
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.091594821913745e-06,
+      "loss": 0.1527,
+      "step": 7207
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.0878610051216525e-06,
+      "loss": 0.1385,
+      "step": 7208
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.084128278263967e-06,
+      "loss": 0.0591,
+      "step": 7209
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.080396641751145e-06,
+      "loss": 0.0886,
+      "step": 7210
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.076666095993508e-06,
+      "loss": 0.1058,
+      "step": 7211
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.072936641401269e-06,
+      "loss": 0.119,
+      "step": 7212
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.06920827838452e-06,
+      "loss": 0.1043,
+      "step": 7213
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.065481007353228e-06,
+      "loss": 0.0706,
+      "step": 7214
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.061754828717237e-06,
+      "loss": 0.0723,
+      "step": 7215
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.05802974288628e-06,
+      "loss": 0.1162,
+      "step": 7216
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.0543057502699654e-06,
+      "loss": 0.0882,
+      "step": 7217
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.050582851277787e-06,
+      "loss": 0.0988,
+      "step": 7218
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.046861046319109e-06,
+      "loss": 0.0345,
+      "step": 7219
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.043140335803174e-06,
+      "loss": 0.1118,
+      "step": 7220
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.0394207201391135e-06,
+      "loss": 0.0604,
+      "step": 7221
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.035702199735941e-06,
+      "loss": 0.0783,
+      "step": 7222
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.031984775002534e-06,
+      "loss": 0.1635,
+      "step": 7223
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.028268446347664e-06,
+      "loss": 0.1251,
+      "step": 7224
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.02455321417998e-06,
+      "loss": 0.1099,
+      "step": 7225
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.020839078908003e-06,
+      "loss": 0.0993,
+      "step": 7226
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.017126040940132e-06,
+      "loss": 0.0855,
+      "step": 7227
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.0134141006846585e-06,
+      "loss": 0.0548,
+      "step": 7228
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.009703258549744e-06,
+      "loss": 0.1045,
+      "step": 7229
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.005993514943435e-06,
+      "loss": 0.1117,
+      "step": 7230
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.002284870273651e-06,
+      "loss": 0.0869,
+      "step": 7231
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.998577324948185e-06,
+      "loss": 0.1281,
+      "step": 7232
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.994870879374724e-06,
+      "loss": 0.0762,
+      "step": 7233
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.991165533960828e-06,
+      "loss": 0.0881,
+      "step": 7234
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.987461289113937e-06,
+      "loss": 0.1078,
+      "step": 7235
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.983758145241365e-06,
+      "loss": 0.0688,
+      "step": 7236
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.980056102750305e-06,
+      "loss": 0.0916,
+      "step": 7237
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.976355162047833e-06,
+      "loss": 0.0811,
+      "step": 7238
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.97265532354091e-06,
+      "loss": 0.123,
+      "step": 7239
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.968956587636357e-06,
+      "loss": 0.0841,
+      "step": 7240
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.9652589547408986e-06,
+      "loss": 0.082,
+      "step": 7241
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.961562425261112e-06,
+      "loss": 0.1063,
+      "step": 7242
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.957866999603476e-06,
+      "loss": 0.068,
+      "step": 7243
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.9541726781743305e-06,
+      "loss": 0.0848,
+      "step": 7244
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.950479461379903e-06,
+      "loss": 0.1115,
+      "step": 7245
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.946787349626302e-06,
+      "loss": 0.1349,
+      "step": 7246
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.943096343319511e-06,
+      "loss": 0.0757,
+      "step": 7247
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.93940644286539e-06,
+      "loss": 0.1217,
+      "step": 7248
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.9357176486696715e-06,
+      "loss": 0.0963,
+      "step": 7249
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.93202996113798e-06,
+      "loss": 0.0954,
+      "step": 7250
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.928343380675817e-06,
+      "loss": 0.0998,
+      "step": 7251
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.9246579076885475e-06,
+      "loss": 0.0683,
+      "step": 7252
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.920973542581432e-06,
+      "loss": 0.1211,
+      "step": 7253
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.917290285759594e-06,
+      "loss": 0.062,
+      "step": 7254
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.913608137628054e-06,
+      "loss": 0.0954,
+      "step": 7255
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.909927098591686e-06,
+      "loss": 0.0695,
+      "step": 7256
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.906247169055262e-06,
+      "loss": 0.0844,
+      "step": 7257
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.902568349423431e-06,
+      "loss": 0.0862,
+      "step": 7258
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.8988906401007086e-06,
+      "loss": 0.1589,
+      "step": 7259
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.89521404149149e-06,
+      "loss": 0.0714,
+      "step": 7260
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.891538554000057e-06,
+      "loss": 0.0926,
+      "step": 7261
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.887864178030563e-06,
+      "loss": 0.0548,
+      "step": 7262
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.884190913987048e-06,
+      "loss": 0.0865,
+      "step": 7263
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.880518762273416e-06,
+      "loss": 0.1102,
+      "step": 7264
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.87684772329345e-06,
+      "loss": 0.1294,
+      "step": 7265
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.8731777974508214e-06,
+      "loss": 0.0691,
+      "step": 7266
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.8695089851490765e-06,
+      "loss": 0.1161,
+      "step": 7267
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.865841286791636e-06,
+      "loss": 0.0988,
+      "step": 7268
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.862174702781797e-06,
+      "loss": 0.076,
+      "step": 7269
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.8585092335227295e-06,
+      "loss": 0.0672,
+      "step": 7270
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.854844879417491e-06,
+      "loss": 0.0879,
+      "step": 7271
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.851181640869018e-06,
+      "loss": 0.1255,
+      "step": 7272
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.847519518280108e-06,
+      "loss": 0.0882,
+      "step": 7273
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.843858512053453e-06,
+      "loss": 0.1123,
+      "step": 7274
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.840198622591619e-06,
+      "loss": 0.05,
+      "step": 7275
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.836539850297042e-06,
+      "loss": 0.0765,
+      "step": 7276
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.832882195572033e-06,
+      "loss": 0.1052,
+      "step": 7277
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.829225658818792e-06,
+      "loss": 0.0886,
+      "step": 7278
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.825570240439388e-06,
+      "loss": 0.0394,
+      "step": 7279
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.821915940835778e-06,
+      "loss": 0.1302,
+      "step": 7280
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.818262760409778e-06,
+      "loss": 0.0801,
+      "step": 7281
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.814610699563089e-06,
+      "loss": 0.0899,
+      "step": 7282
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.810959758697291e-06,
+      "loss": 0.1228,
+      "step": 7283
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.8073099382138495e-06,
+      "loss": 0.1386,
+      "step": 7284
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.8036612385140826e-06,
+      "loss": 0.0562,
+      "step": 7285
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.800013659999211e-06,
+      "loss": 0.115,
+      "step": 7286
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.796367203070314e-06,
+      "loss": 0.1539,
+      "step": 7287
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.792721868128359e-06,
+      "loss": 0.1182,
+      "step": 7288
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.78907765557418e-06,
+      "loss": 0.1208,
+      "step": 7289
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.785434565808496e-06,
+      "loss": 0.1109,
+      "step": 7290
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.781792599231905e-06,
+      "loss": 0.0989,
+      "step": 7291
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.778151756244871e-06,
+      "loss": 0.0812,
+      "step": 7292
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.774512037247733e-06,
+      "loss": 0.1097,
+      "step": 7293
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.770873442640721e-06,
+      "loss": 0.0875,
+      "step": 7294
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7672359728239305e-06,
+      "loss": 0.0747,
+      "step": 7295
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.763599628197337e-06,
+      "loss": 0.0509,
+      "step": 7296
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.759964409160802e-06,
+      "loss": 0.063,
+      "step": 7297
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.75633031611403e-06,
+      "loss": 0.0969,
+      "step": 7298
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.752697349456638e-06,
+      "loss": 0.127,
+      "step": 7299
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.749065509588103e-06,
+      "loss": 0.115,
+      "step": 7300
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.745434796907786e-06,
+      "loss": 0.0836,
+      "step": 7301
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.741805211814909e-06,
+      "loss": 0.1115,
+      "step": 7302
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.738176754708589e-06,
+      "loss": 0.0822,
+      "step": 7303
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.734549425987801e-06,
+      "loss": 0.077,
+      "step": 7304
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.730923226051412e-06,
+      "loss": 0.0945,
+      "step": 7305
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.727298155298148e-06,
+      "loss": 0.0588,
+      "step": 7306
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.723674214126627e-06,
+      "loss": 0.0768,
+      "step": 7307
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.720051402935338e-06,
+      "loss": 0.086,
+      "step": 7308
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.716429722122643e-06,
+      "loss": 0.0717,
+      "step": 7309
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.712809172086771e-06,
+      "loss": 0.0871,
+      "step": 7310
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.709189753225843e-06,
+      "loss": 0.1204,
+      "step": 7311
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.705571465937849e-06,
+      "loss": 0.0811,
+      "step": 7312
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.701954310620659e-06,
+      "loss": 0.0879,
+      "step": 7313
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.698338287672009e-06,
+      "loss": 0.0686,
+      "step": 7314
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.694723397489508e-06,
+      "loss": 0.1222,
+      "step": 7315
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.691109640470657e-06,
+      "loss": 0.0978,
+      "step": 7316
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.687497017012824e-06,
+      "loss": 0.0712,
+      "step": 7317
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.683885527513244e-06,
+      "loss": 0.0632,
+      "step": 7318
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.6802751723690455e-06,
+      "loss": 0.0966,
+      "step": 7319
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.676665951977208e-06,
+      "loss": 0.1276,
+      "step": 7320
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.673057866734613e-06,
+      "loss": 0.0414,
+      "step": 7321
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.6694509170379925e-06,
+      "loss": 0.1067,
+      "step": 7322
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.6658451032839704e-06,
+      "loss": 0.0859,
+      "step": 7323
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.66224042586904e-06,
+      "loss": 0.1525,
+      "step": 7324
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.658636885189577e-06,
+      "loss": 0.0306,
+      "step": 7325
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.655034481641816e-06,
+      "loss": 0.105,
+      "step": 7326
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.651433215621875e-06,
+      "loss": 0.093,
+      "step": 7327
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.647833087525751e-06,
+      "loss": 0.1122,
+      "step": 7328
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.644234097749312e-06,
+      "loss": 0.1228,
+      "step": 7329
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.640636246688305e-06,
+      "loss": 0.0517,
+      "step": 7330
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.637039534738347e-06,
+      "loss": 0.1012,
+      "step": 7331
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.633443962294923e-06,
+      "loss": 0.129,
+      "step": 7332
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.6298495297534075e-06,
+      "loss": 0.0468,
+      "step": 7333
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.626256237509044e-06,
+      "loss": 0.1426,
+      "step": 7334
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.622664085956942e-06,
+      "loss": 0.0796,
+      "step": 7335
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.6190730754921035e-06,
+      "loss": 0.0904,
+      "step": 7336
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.615483206509385e-06,
+      "loss": 0.0739,
+      "step": 7337
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.611894479403533e-06,
+      "loss": 0.099,
+      "step": 7338
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.608306894569156e-06,
+      "loss": 0.0627,
+      "step": 7339
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.604720452400747e-06,
+      "loss": 0.068,
+      "step": 7340
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.601135153292673e-06,
+      "loss": 0.0826,
+      "step": 7341
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.59755099763917e-06,
+      "loss": 0.1481,
+      "step": 7342
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.593967985834343e-06,
+      "loss": 0.1229,
+      "step": 7343
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.590386118272186e-06,
+      "loss": 0.0625,
+      "step": 7344
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.586805395346556e-06,
+      "loss": 0.1208,
+      "step": 7345
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.5832258174511945e-06,
+      "loss": 0.0839,
+      "step": 7346
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.579647384979706e-06,
+      "loss": 0.05,
+      "step": 7347
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.576070098325566e-06,
+      "loss": 0.0619,
+      "step": 7348
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.57249395788214e-06,
+      "loss": 0.1136,
+      "step": 7349
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.5689189640426585e-06,
+      "loss": 0.0587,
+      "step": 7350
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.565345117200223e-06,
+      "loss": 0.066,
+      "step": 7351
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.56177241774781e-06,
+      "loss": 0.0967,
+      "step": 7352
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.558200866078283e-06,
+      "loss": 0.0865,
+      "step": 7353
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.5546304625843606e-06,
+      "loss": 0.1111,
+      "step": 7354
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.551061207658637e-06,
+      "loss": 0.0495,
+      "step": 7355
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.5474931016935924e-06,
+      "loss": 0.0587,
+      "step": 7356
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.543926145081574e-06,
+      "loss": 0.0351,
+      "step": 7357
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.540360338214805e-06,
+      "loss": 0.0733,
+      "step": 7358
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.53679568148538e-06,
+      "loss": 0.144,
+      "step": 7359
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.5332321752852566e-06,
+      "loss": 0.1298,
+      "step": 7360
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.529669820006286e-06,
+      "loss": 0.0905,
+      "step": 7361
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.526108616040181e-06,
+      "loss": 0.0653,
+      "step": 7362
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.5225485637785355e-06,
+      "loss": 0.0238,
+      "step": 7363
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.518989663612804e-06,
+      "loss": 0.1522,
+      "step": 7364
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.51543191593432e-06,
+      "loss": 0.1284,
+      "step": 7365
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.511875321134298e-06,
+      "loss": 0.0722,
+      "step": 7366
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.508319879603818e-06,
+      "loss": 0.0849,
+      "step": 7367
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.504765591733833e-06,
+      "loss": 0.0643,
+      "step": 7368
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.501212457915173e-06,
+      "loss": 0.1338,
+      "step": 7369
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.497660478538535e-06,
+      "loss": 0.022,
+      "step": 7370
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.494109653994499e-06,
+      "loss": 0.1218,
+      "step": 7371
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.490559984673507e-06,
+      "loss": 0.0818,
+      "step": 7372
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.487011470965879e-06,
+      "loss": 0.0924,
+      "step": 7373
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4834641132618095e-06,
+      "loss": 0.0994,
+      "step": 7374
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.479917911951376e-06,
+      "loss": 0.097,
+      "step": 7375
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4763728674244965e-06,
+      "loss": 0.0698,
+      "step": 7376
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.472828980070992e-06,
+      "loss": 0.0846,
+      "step": 7377
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.469286250280546e-06,
+      "loss": 0.0992,
+      "step": 7378
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.465744678442723e-06,
+      "loss": 0.113,
+      "step": 7379
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.46220426494694e-06,
+      "loss": 0.0952,
+      "step": 7380
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.45866501018251e-06,
+      "loss": 0.0446,
+      "step": 7381
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.455126914538599e-06,
+      "loss": 0.0902,
+      "step": 7382
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.451589978404263e-06,
+      "loss": 0.0859,
+      "step": 7383
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.448054202168413e-06,
+      "loss": 0.0634,
+      "step": 7384
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4445195862198475e-06,
+      "loss": 0.0594,
+      "step": 7385
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4409861309472325e-06,
+      "loss": 0.1141,
+      "step": 7386
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.437453836739101e-06,
+      "loss": 0.0716,
+      "step": 7387
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.433922703983865e-06,
+      "loss": 0.0356,
+      "step": 7388
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.430392733069804e-06,
+      "loss": 0.1122,
+      "step": 7389
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4268639243850696e-06,
+      "loss": 0.0806,
+      "step": 7390
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.423336278317699e-06,
+      "loss": 0.0876,
+      "step": 7391
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.419809795255583e-06,
+      "loss": 0.0869,
+      "step": 7392
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.416284475586487e-06,
+      "loss": 0.0758,
+      "step": 7393
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.412760319698061e-06,
+      "loss": 0.1223,
+      "step": 7394
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.409237327977816e-06,
+      "loss": 0.122,
+      "step": 7395
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.405715500813146e-06,
+      "loss": 0.0869,
+      "step": 7396
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4021948385913045e-06,
+      "loss": 0.0666,
+      "step": 7397
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.398675341699416e-06,
+      "loss": 0.087,
+      "step": 7398
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.395157010524486e-06,
+      "loss": 0.0626,
+      "step": 7399
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.391639845453398e-06,
+      "loss": 0.0768,
+      "step": 7400
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.388123846872885e-06,
+      "loss": 0.0978,
+      "step": 7401
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.3846090151695714e-06,
+      "loss": 0.09,
+      "step": 7402
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.381095350729949e-06,
+      "loss": 0.1102,
+      "step": 7403
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.377582853940376e-06,
+      "loss": 0.0855,
+      "step": 7404
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.374071525187079e-06,
+      "loss": 0.0927,
+      "step": 7405
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.370561364856168e-06,
+      "loss": 0.1193,
+      "step": 7406
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.3670523733336175e-06,
+      "loss": 0.084,
+      "step": 7407
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.3635445510052806e-06,
+      "loss": 0.0779,
+      "step": 7408
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.360037898256871e-06,
+      "loss": 0.1087,
+      "step": 7409
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.356532415473972e-06,
+      "loss": 0.1237,
+      "step": 7410
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.353028103042052e-06,
+      "loss": 0.0934,
+      "step": 7411
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.349524961346447e-06,
+      "loss": 0.1087,
+      "step": 7412
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.346022990772354e-06,
+      "loss": 0.0913,
+      "step": 7413
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.3425221917048534e-06,
+      "loss": 0.1021,
+      "step": 7414
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.3390225645288835e-06,
+      "loss": 0.0636,
+      "step": 7415
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.335524109629269e-06,
+      "loss": 0.1246,
+      "step": 7416
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.332026827390702e-06,
+      "loss": 0.0866,
+      "step": 7417
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.3285307181977305e-06,
+      "loss": 0.0826,
+      "step": 7418
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.325035782434797e-06,
+      "loss": 0.0798,
+      "step": 7419
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.321542020486191e-06,
+      "loss": 0.0966,
+      "step": 7420
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.3180494327360985e-06,
+      "loss": 0.1337,
+      "step": 7421
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.314558019568549e-06,
+      "loss": 0.0822,
+      "step": 7422
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.311067781367466e-06,
+      "loss": 0.0704,
+      "step": 7423
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.30757871851663e-06,
+      "loss": 0.0612,
+      "step": 7424
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.304090831399709e-06,
+      "loss": 0.0654,
+      "step": 7425
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.300604120400211e-06,
+      "loss": 0.0868,
+      "step": 7426
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.297118585901542e-06,
+      "loss": 0.0554,
+      "step": 7427
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.293634228286967e-06,
+      "loss": 0.1118,
+      "step": 7428
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.290151047939633e-06,
+      "loss": 0.0555,
+      "step": 7429
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.28666904524254e-06,
+      "loss": 0.0535,
+      "step": 7430
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.283188220578573e-06,
+      "loss": 0.0711,
+      "step": 7431
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.279708574330476e-06,
+      "loss": 0.0774,
+      "step": 7432
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.276230106880875e-06,
+      "loss": 0.0848,
+      "step": 7433
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.272752818612256e-06,
+      "loss": 0.0913,
+      "step": 7434
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.269276709906983e-06,
+      "loss": 0.0611,
+      "step": 7435
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.26580178114729e-06,
+      "loss": 0.0954,
+      "step": 7436
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.262328032715274e-06,
+      "loss": 0.0686,
+      "step": 7437
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.258855464992907e-06,
+      "loss": 0.0729,
+      "step": 7438
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.25538407836203e-06,
+      "loss": 0.0804,
+      "step": 7439
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2519138732043574e-06,
+      "loss": 0.0794,
+      "step": 7440
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.248444849901479e-06,
+      "loss": 0.0782,
+      "step": 7441
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.244977008834837e-06,
+      "loss": 0.0929,
+      "step": 7442
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.241510350385751e-06,
+      "loss": 0.0762,
+      "step": 7443
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.238044874935421e-06,
+      "loss": 0.1402,
+      "step": 7444
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.234580582864905e-06,
+      "loss": 0.0791,
+      "step": 7445
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.231117474555141e-06,
+      "loss": 0.0726,
+      "step": 7446
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2276555503869264e-06,
+      "loss": 0.1058,
+      "step": 7447
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.22419481074093e-06,
+      "loss": 0.088,
+      "step": 7448
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2207352559976945e-06,
+      "loss": 0.0612,
+      "step": 7449
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2172768865376386e-06,
+      "loss": 0.1021,
+      "step": 7450
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.21381970274103e-06,
+      "loss": 0.065,
+      "step": 7451
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2103637049880275e-06,
+      "loss": 0.0543,
+      "step": 7452
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.206908893658656e-06,
+      "loss": 0.0868,
+      "step": 7453
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.203455269132797e-06,
+      "loss": 0.1216,
+      "step": 7454
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.200002831790207e-06,
+      "loss": 0.1137,
+      "step": 7455
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.196551582010519e-06,
+      "loss": 0.1253,
+      "step": 7456
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.193101520173231e-06,
+      "loss": 0.104,
+      "step": 7457
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.189652646657714e-06,
+      "loss": 0.0556,
+      "step": 7458
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.1862049618432e-06,
+      "loss": 0.0992,
+      "step": 7459
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.182758466108791e-06,
+      "loss": 0.0608,
+      "step": 7460
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.179313159833468e-06,
+      "loss": 0.0935,
+      "step": 7461
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.175869043396079e-06,
+      "loss": 0.1234,
+      "step": 7462
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.1724261171753255e-06,
+      "loss": 0.1063,
+      "step": 7463
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.1689843815498025e-06,
+      "loss": 0.1139,
+      "step": 7464
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.165543836897953e-06,
+      "loss": 0.0955,
+      "step": 7465
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.162104483598105e-06,
+      "loss": 0.1414,
+      "step": 7466
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.15866632202844e-06,
+      "loss": 0.0904,
+      "step": 7467
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.15522935256702e-06,
+      "loss": 0.1186,
+      "step": 7468
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.1517935755917794e-06,
+      "loss": 0.0785,
+      "step": 7469
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.148358991480509e-06,
+      "loss": 0.1187,
+      "step": 7470
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.144925600610869e-06,
+      "loss": 0.0657,
+      "step": 7471
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.141493403360399e-06,
+      "loss": 0.0738,
+      "step": 7472
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.138062400106501e-06,
+      "loss": 0.0601,
+      "step": 7473
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.134632591226453e-06,
+      "loss": 0.0689,
+      "step": 7474
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.1312039770973875e-06,
+      "loss": 0.1124,
+      "step": 7475
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.127776558096312e-06,
+      "loss": 0.1258,
+      "step": 7476
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.124350334600109e-06,
+      "loss": 0.0878,
+      "step": 7477
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.120925306985521e-06,
+      "loss": 0.0651,
+      "step": 7478
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.117501475629169e-06,
+      "loss": 0.0541,
+      "step": 7479
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.114078840907528e-06,
+      "loss": 0.1495,
+      "step": 7480
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.1106574031969555e-06,
+      "loss": 0.0816,
+      "step": 7481
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.1072371628736655e-06,
+      "loss": 0.1026,
+      "step": 7482
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.103818120313753e-06,
+      "loss": 0.0853,
+      "step": 7483
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.100400275893166e-06,
+      "loss": 0.0239,
+      "step": 7484
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0969836299877325e-06,
+      "loss": 0.0634,
+      "step": 7485
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.093568182973153e-06,
+      "loss": 0.1112,
+      "step": 7486
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.090153935224981e-06,
+      "loss": 0.0852,
+      "step": 7487
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.086740887118644e-06,
+      "loss": 0.1377,
+      "step": 7488
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.08332903902944e-06,
+      "loss": 0.1059,
+      "step": 7489
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.079918391332537e-06,
+      "loss": 0.0879,
+      "step": 7490
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.076508944402973e-06,
+      "loss": 0.1112,
+      "step": 7491
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.073100698615642e-06,
+      "loss": 0.0467,
+      "step": 7492
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.069693654345313e-06,
+      "loss": 0.0557,
+      "step": 7493
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.066287811966623e-06,
+      "loss": 0.099,
+      "step": 7494
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.062883171854085e-06,
+      "loss": 0.0891,
+      "step": 7495
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.05947973438206e-06,
+      "loss": 0.1158,
+      "step": 7496
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0560774999248005e-06,
+      "loss": 0.0928,
+      "step": 7497
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0526764688564045e-06,
+      "loss": 0.0944,
+      "step": 7498
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.049276641550854e-06,
+      "loss": 0.1373,
+      "step": 7499
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0458780183819875e-06,
+      "loss": 0.1485,
+      "step": 7500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.042480599723517e-06,
+      "loss": 0.0652,
+      "step": 7501
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0390843859490255e-06,
+      "loss": 0.0854,
+      "step": 7502
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.035689377431965e-06,
+      "loss": 0.1205,
+      "step": 7503
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.032295574545631e-06,
+      "loss": 0.0878,
+      "step": 7504
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.028902977663216e-06,
+      "loss": 0.0623,
+      "step": 7505
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.025511587157766e-06,
+      "loss": 0.0939,
+      "step": 7506
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.022121403402198e-06,
+      "loss": 0.0639,
+      "step": 7507
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0187324267693e-06,
+      "loss": 0.0753,
+      "step": 7508
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0153446576317155e-06,
+      "loss": 0.0736,
+      "step": 7509
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.01195809636196e-06,
+      "loss": 0.1139,
+      "step": 7510
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.008572743332422e-06,
+      "loss": 0.1379,
+      "step": 7511
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.005188598915359e-06,
+      "loss": 0.1022,
+      "step": 7512
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0018056634828785e-06,
+      "loss": 0.1186,
+      "step": 7513
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.998423937406979e-06,
+      "loss": 0.1248,
+      "step": 7514
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.995043421059501e-06,
+      "loss": 0.099,
+      "step": 7515
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.991664114812175e-06,
+      "loss": 0.0766,
+      "step": 7516
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.988286019036581e-06,
+      "loss": 0.1117,
+      "step": 7517
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.984909134104175e-06,
+      "loss": 0.0775,
+      "step": 7518
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.9815334603862815e-06,
+      "loss": 0.1073,
+      "step": 7519
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.978158998254086e-06,
+      "loss": 0.0729,
+      "step": 7520
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.974785748078638e-06,
+      "loss": 0.1196,
+      "step": 7521
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.971413710230862e-06,
+      "loss": 0.0868,
+      "step": 7522
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.968042885081547e-06,
+      "loss": 0.1011,
+      "step": 7523
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.9646732730013504e-06,
+      "loss": 0.115,
+      "step": 7524
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.9613048743607905e-06,
+      "loss": 0.061,
+      "step": 7525
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.957937689530248e-06,
+      "loss": 0.1154,
+      "step": 7526
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.954571718879984e-06,
+      "loss": 0.0522,
+      "step": 7527
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.951206962780123e-06,
+      "loss": 0.1157,
+      "step": 7528
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.947843421600642e-06,
+      "loss": 0.0637,
+      "step": 7529
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.944481095711401e-06,
+      "loss": 0.0516,
+      "step": 7530
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.941119985482122e-06,
+      "loss": 0.0992,
+      "step": 7531
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.937760091282387e-06,
+      "loss": 0.0751,
+      "step": 7532
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.934401413481646e-06,
+      "loss": 0.0809,
+      "step": 7533
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.931043952449222e-06,
+      "loss": 0.0862,
+      "step": 7534
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.927687708554297e-06,
+      "loss": 0.0639,
+      "step": 7535
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.92433268216593e-06,
+      "loss": 0.0959,
+      "step": 7536
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.920978873653031e-06,
+      "loss": 0.1167,
+      "step": 7537
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.917626283384381e-06,
+      "loss": 0.1142,
+      "step": 7538
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.914274911728634e-06,
+      "loss": 0.1183,
+      "step": 7539
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.910924759054301e-06,
+      "loss": 0.1453,
+      "step": 7540
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.907575825729774e-06,
+      "loss": 0.088,
+      "step": 7541
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.904228112123292e-06,
+      "loss": 0.1131,
+      "step": 7542
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.900881618602965e-06,
+      "loss": 0.1301,
+      "step": 7543
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.897536345536774e-06,
+      "loss": 0.1432,
+      "step": 7544
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.894192293292572e-06,
+      "loss": 0.0735,
+      "step": 7545
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.890849462238057e-06,
+      "loss": 0.1139,
+      "step": 7546
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.887507852740816e-06,
+      "loss": 0.0867,
+      "step": 7547
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.884167465168283e-06,
+      "loss": 0.0677,
+      "step": 7548
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.880828299887772e-06,
+      "loss": 0.0517,
+      "step": 7549
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.877490357266448e-06,
+      "loss": 0.1089,
+      "step": 7550
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.874153637671355e-06,
+      "loss": 0.1335,
+      "step": 7551
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.870818141469402e-06,
+      "loss": 0.0412,
+      "step": 7552
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.867483869027353e-06,
+      "loss": 0.0826,
+      "step": 7553
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.864150820711839e-06,
+      "loss": 0.086,
+      "step": 7554
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8608189968893645e-06,
+      "loss": 0.0741,
+      "step": 7555
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.857488397926298e-06,
+      "loss": 0.1262,
+      "step": 7556
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8541590241888705e-06,
+      "loss": 0.079,
+      "step": 7557
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8508308760431735e-06,
+      "loss": 0.0658,
+      "step": 7558
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.847503953855176e-06,
+      "loss": 0.1155,
+      "step": 7559
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.844178257990696e-06,
+      "loss": 0.0991,
+      "step": 7560
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.840853788815436e-06,
+      "loss": 0.056,
+      "step": 7561
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.83753054669494e-06,
+      "loss": 0.0927,
+      "step": 7562
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.834208531994641e-06,
+      "loss": 0.0795,
+      "step": 7563
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.830887745079825e-06,
+      "loss": 0.0645,
+      "step": 7564
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8275681863156434e-06,
+      "loss": 0.141,
+      "step": 7565
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.824249856067108e-06,
+      "loss": 0.0925,
+      "step": 7566
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.820932754699105e-06,
+      "loss": 0.0669,
+      "step": 7567
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.81761688257638e-06,
+      "loss": 0.0681,
+      "step": 7568
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8143022400635513e-06,
+      "loss": 0.0756,
+      "step": 7569
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.81098882752509e-06,
+      "loss": 0.1248,
+      "step": 7570
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.807676645325334e-06,
+      "loss": 0.0987,
+      "step": 7571
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8043656938284947e-06,
+      "loss": 0.078,
+      "step": 7572
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8010559733986406e-06,
+      "loss": 0.047,
+      "step": 7573
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.797747484399712e-06,
+      "loss": 0.058,
+      "step": 7574
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7944402271955058e-06,
+      "loss": 0.1781,
+      "step": 7575
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7911342021496802e-06,
+      "loss": 0.0821,
+      "step": 7576
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.787829409625769e-06,
+      "loss": 0.07,
+      "step": 7577
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7845258499871737e-06,
+      "loss": 0.1013,
+      "step": 7578
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.781223523597138e-06,
+      "loss": 0.0892,
+      "step": 7579
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.777922430818792e-06,
+      "loss": 0.0753,
+      "step": 7580
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7746225720151267e-06,
+      "loss": 0.1041,
+      "step": 7581
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.771323947548987e-06,
+      "loss": 0.1118,
+      "step": 7582
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7680265577830856e-06,
+      "loss": 0.0985,
+      "step": 7583
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7647304030800052e-06,
+      "loss": 0.0826,
+      "step": 7584
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7614354838021895e-06,
+      "loss": 0.0811,
+      "step": 7585
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7581418003119517e-06,
+      "loss": 0.1005,
+      "step": 7586
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7548493529714595e-06,
+      "loss": 0.1173,
+      "step": 7587
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7515581421427435e-06,
+      "loss": 0.1182,
+      "step": 7588
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.748268168187708e-06,
+      "loss": 0.1034,
+      "step": 7589
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.744979431468123e-06,
+      "loss": 0.0939,
+      "step": 7590
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.741691932345606e-06,
+      "loss": 0.103,
+      "step": 7591
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7384056711816585e-06,
+      "loss": 0.0904,
+      "step": 7592
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7351206483376273e-06,
+      "loss": 0.1038,
+      "step": 7593
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.731836864174742e-06,
+      "loss": 0.1135,
+      "step": 7594
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.7285543190540773e-06,
+      "loss": 0.0943,
+      "step": 7595
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.7252730133365825e-06,
+      "loss": 0.0888,
+      "step": 7596
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.7219929473830745e-06,
+      "loss": 0.0503,
+      "step": 7597
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.718714121554219e-06,
+      "loss": 0.0758,
+      "step": 7598
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.715436536210562e-06,
+      "loss": 0.0664,
+      "step": 7599
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.712160191712499e-06,
+      "loss": 0.0474,
+      "step": 7600
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.708885088420298e-06,
+      "loss": 0.098,
+      "step": 7601
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.705611226694092e-06,
+      "loss": 0.0845,
+      "step": 7602
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.7023386068938707e-06,
+      "loss": 0.0752,
+      "step": 7603
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6990672293794825e-06,
+      "loss": 0.1168,
+      "step": 7604
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.695797094510654e-06,
+      "loss": 0.0605,
+      "step": 7605
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6925282026469674e-06,
+      "loss": 0.0768,
+      "step": 7606
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6892605541478723e-06,
+      "loss": 0.0901,
+      "step": 7607
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.685994149372668e-06,
+      "loss": 0.0469,
+      "step": 7608
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6827289886805364e-06,
+      "loss": 0.0553,
+      "step": 7609
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6794650724305062e-06,
+      "loss": 0.0951,
+      "step": 7610
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.676202400981483e-06,
+      "loss": 0.0967,
+      "step": 7611
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6729409746922205e-06,
+      "loss": 0.0566,
+      "step": 7612
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6696807939213477e-06,
+      "loss": 0.0766,
+      "step": 7613
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.666421859027357e-06,
+      "loss": 0.0987,
+      "step": 7614
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.663164170368596e-06,
+      "loss": 0.1209,
+      "step": 7615
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6599077283032733e-06,
+      "loss": 0.1063,
+      "step": 7616
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6566525331894696e-06,
+      "loss": 0.1025,
+      "step": 7617
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.653398585385128e-06,
+      "loss": 0.0584,
+      "step": 7618
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6501458852480514e-06,
+      "loss": 0.0696,
+      "step": 7619
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.646894433135903e-06,
+      "loss": 0.1277,
+      "step": 7620
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6436442294062065e-06,
+      "loss": 0.066,
+      "step": 7621
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.640395274416359e-06,
+      "loss": 0.0841,
+      "step": 7622
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.637147568523615e-06,
+      "loss": 0.125,
+      "step": 7623
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.633901112085083e-06,
+      "loss": 0.0971,
+      "step": 7624
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.630655905457752e-06,
+      "loss": 0.0318,
+      "step": 7625
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6274119489984536e-06,
+      "loss": 0.0416,
+      "step": 7626
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6241692430638975e-06,
+      "loss": 0.0773,
+      "step": 7627
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6209277880106543e-06,
+      "loss": 0.1405,
+      "step": 7628
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6176875841951436e-06,
+      "loss": 0.1251,
+      "step": 7629
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6144486319736644e-06,
+      "loss": 0.1106,
+      "step": 7630
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6112109317023633e-06,
+      "loss": 0.051,
+      "step": 7631
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6079744837372646e-06,
+      "loss": 0.0759,
+      "step": 7632
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6047392884342384e-06,
+      "loss": 0.1121,
+      "step": 7633
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.60150534614903e-06,
+      "loss": 0.0561,
+      "step": 7634
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.5982726572372397e-06,
+      "loss": 0.0913,
+      "step": 7635
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.5950412220543407e-06,
+      "loss": 0.0558,
+      "step": 7636
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.5918110409556542e-06,
+      "loss": 0.0658,
+      "step": 7637
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.5885821142963637e-06,
+      "loss": 0.1148,
+      "step": 7638
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.5853544424315266e-06,
+      "loss": 0.0954,
+      "step": 7639
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.5821280257160604e-06,
+      "loss": 0.1202,
+      "step": 7640
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.57890286450473e-06,
+      "loss": 0.1149,
+      "step": 7641
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.575678959152185e-06,
+      "loss": 0.0578,
+      "step": 7642
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.5724563100129115e-06,
+      "loss": 0.1228,
+      "step": 7643
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.569234917441283e-06,
+      "loss": 0.0348,
+      "step": 7644
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.566014781791513e-06,
+      "loss": 0.1061,
+      "step": 7645
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5627959034176877e-06,
+      "loss": 0.0991,
+      "step": 7646
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.559578282673761e-06,
+      "loss": 0.053,
+      "step": 7647
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5563619199135365e-06,
+      "loss": 0.0859,
+      "step": 7648
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.553146815490678e-06,
+      "loss": 0.104,
+      "step": 7649
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.549932969758723e-06,
+      "loss": 0.1402,
+      "step": 7650
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5467203830710616e-06,
+      "loss": 0.117,
+      "step": 7651
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5435090557809563e-06,
+      "loss": 0.0872,
+      "step": 7652
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5402989882415183e-06,
+      "loss": 0.092,
+      "step": 7653
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5370901808057184e-06,
+      "loss": 0.0719,
+      "step": 7654
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5338826338264024e-06,
+      "loss": 0.0678,
+      "step": 7655
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5306763476562703e-06,
+      "loss": 0.0849,
+      "step": 7656
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5274713226478867e-06,
+      "loss": 0.0721,
+      "step": 7657
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.524267559153668e-06,
+      "loss": 0.0498,
+      "step": 7658
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.521065057525907e-06,
+      "loss": 0.035,
+      "step": 7659
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.517863818116739e-06,
+      "loss": 0.0451,
+      "step": 7660
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5146638412781805e-06,
+      "loss": 0.0825,
+      "step": 7661
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.511465127362092e-06,
+      "loss": 0.0645,
+      "step": 7662
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.508267676720206e-06,
+      "loss": 0.1117,
+      "step": 7663
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.505071489704117e-06,
+      "loss": 0.1116,
+      "step": 7664
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5018765666652726e-06,
+      "loss": 0.0544,
+      "step": 7665
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4986829079549797e-06,
+      "loss": 0.1901,
+      "step": 7666
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.495490513924418e-06,
+      "loss": 0.102,
+      "step": 7667
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.492299384924618e-06,
+      "loss": 0.0832,
+      "step": 7668
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4891095213064827e-06,
+      "loss": 0.1032,
+      "step": 7669
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.485920923420763e-06,
+      "loss": 0.0542,
+      "step": 7670
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.482733591618071e-06,
+      "loss": 0.1124,
+      "step": 7671
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.479547526248888e-06,
+      "loss": 0.0519,
+      "step": 7672
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4763627276635574e-06,
+      "loss": 0.0443,
+      "step": 7673
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4731791962122694e-06,
+      "loss": 0.1296,
+      "step": 7674
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.469996932245091e-06,
+      "loss": 0.0438,
+      "step": 7675
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.466815936111937e-06,
+      "loss": 0.0762,
+      "step": 7676
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4636362081625937e-06,
+      "loss": 0.1591,
+      "step": 7677
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4604577487466974e-06,
+      "loss": 0.0503,
+      "step": 7678
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4572805582137516e-06,
+      "loss": 0.139,
+      "step": 7679
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4541046369131235e-06,
+      "loss": 0.0651,
+      "step": 7680
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4509299851940316e-06,
+      "loss": 0.1116,
+      "step": 7681
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4477566034055573e-06,
+      "loss": 0.0591,
+      "step": 7682
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4445844918966447e-06,
+      "loss": 0.0854,
+      "step": 7683
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.441413651016101e-06,
+      "loss": 0.1387,
+      "step": 7684
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.438244081112593e-06,
+      "loss": 0.09,
+      "step": 7685
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4350757825346374e-06,
+      "loss": 0.1344,
+      "step": 7686
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.431908755630627e-06,
+      "loss": 0.1182,
+      "step": 7687
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.428743000748797e-06,
+      "loss": 0.099,
+      "step": 7688
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4255785182372595e-06,
+      "loss": 0.0999,
+      "step": 7689
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.422415308443983e-06,
+      "loss": 0.0863,
+      "step": 7690
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.419253371716782e-06,
+      "loss": 0.1198,
+      "step": 7691
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.416092708403351e-06,
+      "loss": 0.1291,
+      "step": 7692
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4129333188512295e-06,
+      "loss": 0.13,
+      "step": 7693
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4097752034078285e-06,
+      "loss": 0.0601,
+      "step": 7694
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.406618362420403e-06,
+      "loss": 0.0859,
+      "step": 7695
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4034627962360853e-06,
+      "loss": 0.0454,
+      "step": 7696
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4003085052018616e-06,
+      "loss": 0.0725,
+      "step": 7697
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3971554896645745e-06,
+      "loss": 0.0974,
+      "step": 7698
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.394003749970923e-06,
+      "loss": 0.0723,
+      "step": 7699
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.390853286467475e-06,
+      "loss": 0.0614,
+      "step": 7700
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3877040995006536e-06,
+      "loss": 0.1286,
+      "step": 7701
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3845561894167482e-06,
+      "loss": 0.0658,
+      "step": 7702
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3814095565618936e-06,
+      "loss": 0.0736,
+      "step": 7703
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.378264201282093e-06,
+      "loss": 0.0776,
+      "step": 7704
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3751201239232082e-06,
+      "loss": 0.0979,
+      "step": 7705
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.371977324830966e-06,
+      "loss": 0.0939,
+      "step": 7706
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.36883580435094e-06,
+      "loss": 0.0772,
+      "step": 7707
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3656955628285787e-06,
+      "loss": 0.0779,
+      "step": 7708
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3625566006091708e-06,
+      "loss": 0.0572,
+      "step": 7709
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3594189180378854e-06,
+      "loss": 0.133,
+      "step": 7710
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3562825154597332e-06,
+      "loss": 0.0687,
+      "step": 7711
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.353147393219595e-06,
+      "loss": 0.063,
+      "step": 7712
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3500135516622064e-06,
+      "loss": 0.0766,
+      "step": 7713
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3468809911321675e-06,
+      "loss": 0.1195,
+      "step": 7714
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.343749711973932e-06,
+      "loss": 0.0821,
+      "step": 7715
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3406197145318062e-06,
+      "loss": 0.1129,
+      "step": 7716
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3374909991499688e-06,
+      "loss": 0.1043,
+      "step": 7717
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3343635661724538e-06,
+      "loss": 0.056,
+      "step": 7718
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.331237415943153e-06,
+      "loss": 0.0686,
+      "step": 7719
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.328112548805815e-06,
+      "loss": 0.0761,
+      "step": 7720
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.324988965104043e-06,
+      "loss": 0.0679,
+      "step": 7721
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3218666651813095e-06,
+      "loss": 0.067,
+      "step": 7722
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.318745649380946e-06,
+      "loss": 0.0952,
+      "step": 7723
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3156259180461297e-06,
+      "loss": 0.0945,
+      "step": 7724
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3125074715199128e-06,
+      "loss": 0.0571,
+      "step": 7725
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3093903101451893e-06,
+      "loss": 0.1479,
+      "step": 7726
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3062744342647306e-06,
+      "loss": 0.0408,
+      "step": 7727
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3031598442211468e-06,
+      "loss": 0.1436,
+      "step": 7728
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3000465403569244e-06,
+      "loss": 0.0706,
+      "step": 7729
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2969345230144014e-06,
+      "loss": 0.1439,
+      "step": 7730
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2938237925357704e-06,
+      "loss": 0.0746,
+      "step": 7731
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2907143492630846e-06,
+      "loss": 0.0859,
+      "step": 7732
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2876061935382572e-06,
+      "loss": 0.0775,
+      "step": 7733
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2844993257030632e-06,
+      "loss": 0.0679,
+      "step": 7734
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2813937460991338e-06,
+      "loss": 0.0562,
+      "step": 7735
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2782894550679515e-06,
+      "loss": 0.0751,
+      "step": 7736
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2751864529508677e-06,
+      "loss": 0.0898,
+      "step": 7737
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2720847400890817e-06,
+      "loss": 0.0692,
+      "step": 7738
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2689843168236623e-06,
+      "loss": 0.0927,
+      "step": 7739
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2658851834955262e-06,
+      "loss": 0.0954,
+      "step": 7740
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2627873404454527e-06,
+      "loss": 0.0786,
+      "step": 7741
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.259690788014085e-06,
+      "loss": 0.1051,
+      "step": 7742
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.256595526541916e-06,
+      "loss": 0.1568,
+      "step": 7743
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2535015563692932e-06,
+      "loss": 0.0757,
+      "step": 7744
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.250408877836432e-06,
+      "loss": 0.0947,
+      "step": 7745
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2473174912834044e-06,
+      "loss": 0.0867,
+      "step": 7746
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2442273970501392e-06,
+      "loss": 0.0825,
+      "step": 7747
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.24113859547642e-06,
+      "loss": 0.0818,
+      "step": 7748
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2380510869018836e-06,
+      "loss": 0.0728,
+      "step": 7749
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2349648716660386e-06,
+      "loss": 0.0695,
+      "step": 7750
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2318799501082396e-06,
+      "loss": 0.092,
+      "step": 7751
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.228796322567711e-06,
+      "loss": 0.0842,
+      "step": 7752
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.225713989383521e-06,
+      "loss": 0.0608,
+      "step": 7753
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2226329508945982e-06,
+      "loss": 0.0695,
+      "step": 7754
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.219553207439736e-06,
+      "loss": 0.0739,
+      "step": 7755
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.216474759357586e-06,
+      "loss": 0.0579,
+      "step": 7756
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2133976069866443e-06,
+      "loss": 0.085,
+      "step": 7757
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2103217506652803e-06,
+      "loss": 0.0716,
+      "step": 7758
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2072471907317084e-06,
+      "loss": 0.075,
+      "step": 7759
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2041739275240133e-06,
+      "loss": 0.092,
+      "step": 7760
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2011019613801212e-06,
+      "loss": 0.1265,
+      "step": 7761
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.198031292637827e-06,
+      "loss": 0.067,
+      "step": 7762
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.19496192163478e-06,
+      "loss": 0.089,
+      "step": 7763
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.191893848708493e-06,
+      "loss": 0.1035,
+      "step": 7764
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1888270741963253e-06,
+      "loss": 0.1502,
+      "step": 7765
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1857615984354926e-06,
+      "loss": 0.0737,
+      "step": 7766
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1826974217630783e-06,
+      "loss": 0.0541,
+      "step": 7767
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1796345445160213e-06,
+      "loss": 0.0514,
+      "step": 7768
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1765729670311074e-06,
+      "loss": 0.0518,
+      "step": 7769
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1735126896449935e-06,
+      "loss": 0.1523,
+      "step": 7770
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.170453712694179e-06,
+      "loss": 0.0605,
+      "step": 7771
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1673960365150333e-06,
+      "loss": 0.1479,
+      "step": 7772
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1643396614437725e-06,
+      "loss": 0.0732,
+      "step": 7773
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.161284587816476e-06,
+      "loss": 0.0913,
+      "step": 7774
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1582308159690846e-06,
+      "loss": 0.0841,
+      "step": 7775
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.155178346237382e-06,
+      "loss": 0.0644,
+      "step": 7776
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.152127178957015e-06,
+      "loss": 0.093,
+      "step": 7777
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1490773144634928e-06,
+      "loss": 0.1228,
+      "step": 7778
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.146028753092177e-06,
+      "loss": 0.0547,
+      "step": 7779
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1429814951782887e-06,
+      "loss": 0.0917,
+      "step": 7780
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1399355410568993e-06,
+      "loss": 0.0811,
+      "step": 7781
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.136890891062939e-06,
+      "loss": 0.0926,
+      "step": 7782
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.133847545531197e-06,
+      "loss": 0.0975,
+      "step": 7783
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.130805504796319e-06,
+      "loss": 0.0881,
+      "step": 7784
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1277647691928126e-06,
+      "loss": 0.1045,
+      "step": 7785
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1247253390550246e-06,
+      "loss": 0.0408,
+      "step": 7786
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.12168721471718e-06,
+      "loss": 0.0507,
+      "step": 7787
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1186503965133396e-06,
+      "loss": 0.1023,
+      "step": 7788
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1156148847774395e-06,
+      "loss": 0.078,
+      "step": 7789
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.1125806798432547e-06,
+      "loss": 0.0774,
+      "step": 7790
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.109547782044432e-06,
+      "loss": 0.0585,
+      "step": 7791
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.106516191714467e-06,
+      "loss": 0.1032,
+      "step": 7792
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.1034859091867102e-06,
+      "loss": 0.0889,
+      "step": 7793
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.1004569347943666e-06,
+      "loss": 0.1051,
+      "step": 7794
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0974292688705034e-06,
+      "loss": 0.0743,
+      "step": 7795
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0944029117480445e-06,
+      "loss": 0.0669,
+      "step": 7796
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.091377863759766e-06,
+      "loss": 0.0784,
+      "step": 7797
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.088354125238301e-06,
+      "loss": 0.0842,
+      "step": 7798
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.085331696516132e-06,
+      "loss": 0.076,
+      "step": 7799
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0823105779256095e-06,
+      "loss": 0.0981,
+      "step": 7800
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0792907697989373e-06,
+      "loss": 0.1131,
+      "step": 7801
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0762722724681643e-06,
+      "loss": 0.0836,
+      "step": 7802
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.073255086265212e-06,
+      "loss": 0.0817,
+      "step": 7803
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.070239211521839e-06,
+      "loss": 0.1035,
+      "step": 7804
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.06722464856968e-06,
+      "loss": 0.0933,
+      "step": 7805
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.064211397740205e-06,
+      "loss": 0.1136,
+      "step": 7806
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0611994593647543e-06,
+      "loss": 0.0643,
+      "step": 7807
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0581888337745234e-06,
+      "loss": 0.0607,
+      "step": 7808
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0551795213005533e-06,
+      "loss": 0.0343,
+      "step": 7809
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0521715222737513e-06,
+      "loss": 0.0304,
+      "step": 7810
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0491648370248687e-06,
+      "loss": 0.0827,
+      "step": 7811
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0461594658845247e-06,
+      "loss": 0.0738,
+      "step": 7812
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.043155409183187e-06,
+      "loss": 0.09,
+      "step": 7813
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0401526672511843e-06,
+      "loss": 0.079,
+      "step": 7814
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0371512404186956e-06,
+      "loss": 0.0548,
+      "step": 7815
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.034151129015749e-06,
+      "loss": 0.0846,
+      "step": 7816
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0311523333722406e-06,
+      "loss": 0.0975,
+      "step": 7817
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.02815485381792e-06,
+      "loss": 0.1058,
+      "step": 7818
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.025158690682383e-06,
+      "loss": 0.04,
+      "step": 7819
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0221638442950916e-06,
+      "loss": 0.0712,
+      "step": 7820
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.019170314985352e-06,
+      "loss": 0.1128,
+      "step": 7821
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0161781030823366e-06,
+      "loss": 0.0678,
+      "step": 7822
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0131872089150632e-06,
+      "loss": 0.1072,
+      "step": 7823
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.010197632812412e-06,
+      "loss": 0.1192,
+      "step": 7824
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.007209375103117e-06,
+      "loss": 0.0919,
+      "step": 7825
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0042224361157657e-06,
+      "loss": 0.0686,
+      "step": 7826
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.001236816178793e-06,
+      "loss": 0.1266,
+      "step": 7827
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.9982525156205016e-06,
+      "loss": 0.072,
+      "step": 7828
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.9952695347690455e-06,
+      "loss": 0.0724,
+      "step": 7829
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.9922878739524327e-06,
+      "loss": 0.0503,
+      "step": 7830
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.9893075334985242e-06,
+      "loss": 0.0753,
+      "step": 7831
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.986328513735033e-06,
+      "loss": 0.1336,
+      "step": 7832
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.983350814989533e-06,
+      "loss": 0.0682,
+      "step": 7833
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.9803744375894543e-06,
+      "loss": 0.0606,
+      "step": 7834
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.977399381862072e-06,
+      "loss": 0.0948,
+      "step": 7835
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.9744256481345276e-06,
+      "loss": 0.0847,
+      "step": 7836
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.9714532367338067e-06,
+      "loss": 0.0877,
+      "step": 7837
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.968482147986755e-06,
+      "loss": 0.0682,
+      "step": 7838
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.965512382220079e-06,
+      "loss": 0.1022,
+      "step": 7839
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9625439397603237e-06,
+      "loss": 0.1337,
+      "step": 7840
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9595768209339015e-06,
+      "loss": 0.0611,
+      "step": 7841
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9566110260670776e-06,
+      "loss": 0.0328,
+      "step": 7842
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.953646555485968e-06,
+      "loss": 0.0896,
+      "step": 7843
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.950683409516541e-06,
+      "loss": 0.0718,
+      "step": 7844
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.947721588484625e-06,
+      "loss": 0.0977,
+      "step": 7845
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9447610927159013e-06,
+      "loss": 0.1024,
+      "step": 7846
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.94180192253591e-06,
+      "loss": 0.1315,
+      "step": 7847
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.938844078270034e-06,
+      "loss": 0.0595,
+      "step": 7848
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9358875602435126e-06,
+      "loss": 0.0848,
+      "step": 7849
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.932932368781448e-06,
+      "loss": 0.0833,
+      "step": 7850
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.929978504208795e-06,
+      "loss": 0.0753,
+      "step": 7851
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9270259668503542e-06,
+      "loss": 0.0769,
+      "step": 7852
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9240747570307904e-06,
+      "loss": 0.0877,
+      "step": 7853
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9211248750746088e-06,
+      "loss": 0.0836,
+      "step": 7854
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9181763213061874e-06,
+      "loss": 0.0942,
+      "step": 7855
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.915229096049738e-06,
+      "loss": 0.1096,
+      "step": 7856
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9122831996293414e-06,
+      "loss": 0.0914,
+      "step": 7857
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.90933863236893e-06,
+      "loss": 0.0623,
+      "step": 7858
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.906395394592284e-06,
+      "loss": 0.0553,
+      "step": 7859
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9034534866230366e-06,
+      "loss": 0.1294,
+      "step": 7860
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9005129087846814e-06,
+      "loss": 0.08,
+      "step": 7861
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.897573661400565e-06,
+      "loss": 0.143,
+      "step": 7862
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.89463574479389e-06,
+      "loss": 0.1578,
+      "step": 7863
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8916991592876962e-06,
+      "loss": 0.1354,
+      "step": 7864
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8887639052049024e-06,
+      "loss": 0.1004,
+      "step": 7865
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8858299828682555e-06,
+      "loss": 0.0965,
+      "step": 7866
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8828973926003806e-06,
+      "loss": 0.0722,
+      "step": 7867
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8799661347237326e-06,
+      "loss": 0.0972,
+      "step": 7868
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.877036209560636e-06,
+      "loss": 0.1222,
+      "step": 7869
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8741076174332698e-06,
+      "loss": 0.0772,
+      "step": 7870
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.871180358663651e-06,
+      "loss": 0.0736,
+      "step": 7871
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8682544335736686e-06,
+      "loss": 0.1036,
+      "step": 7872
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.865329842485048e-06,
+      "loss": 0.079,
+      "step": 7873
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.862406585719377e-06,
+      "loss": 0.0697,
+      "step": 7874
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8594846635981046e-06,
+      "loss": 0.1429,
+      "step": 7875
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.856564076442517e-06,
+      "loss": 0.1312,
+      "step": 7876
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8536448245737573e-06,
+      "loss": 0.0625,
+      "step": 7877
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8507269083128286e-06,
+      "loss": 0.0988,
+      "step": 7878
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8478103279805854e-06,
+      "loss": 0.0835,
+      "step": 7879
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8448950838977356e-06,
+      "loss": 0.1137,
+      "step": 7880
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8419811763848354e-06,
+      "loss": 0.0871,
+      "step": 7881
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8390686057622915e-06,
+      "loss": 0.1323,
+      "step": 7882
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8361573723503735e-06,
+      "loss": 0.1052,
+      "step": 7883
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.833247476469205e-06,
+      "loss": 0.1006,
+      "step": 7884
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8303389184387467e-06,
+      "loss": 0.0551,
+      "step": 7885
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8274316985788312e-06,
+      "loss": 0.1325,
+      "step": 7886
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.824525817209126e-06,
+      "loss": 0.084,
+      "step": 7887
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.8216212746491704e-06,
+      "loss": 0.1059,
+      "step": 7888
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.8187180712183374e-06,
+      "loss": 0.0299,
+      "step": 7889
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.8158162072358667e-06,
+      "loss": 0.0994,
+      "step": 7890
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.8129156830208447e-06,
+      "loss": 0.0626,
+      "step": 7891
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.810016498892216e-06,
+      "loss": 0.0809,
+      "step": 7892
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.807118655168771e-06,
+      "loss": 0.0713,
+      "step": 7893
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.8042221521691513e-06,
+      "loss": 0.1006,
+      "step": 7894
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.8013269902118587e-06,
+      "loss": 0.0802,
+      "step": 7895
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7984331696152455e-06,
+      "loss": 0.0526,
+      "step": 7896
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.79554069069751e-06,
+      "loss": 0.089,
+      "step": 7897
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.792649553776715e-06,
+      "loss": 0.0565,
+      "step": 7898
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7897597591707608e-06,
+      "loss": 0.0867,
+      "step": 7899
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.78687130719741e-06,
+      "loss": 0.0816,
+      "step": 7900
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.783984198174282e-06,
+      "loss": 0.1271,
+      "step": 7901
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7810984324188343e-06,
+      "loss": 0.0976,
+      "step": 7902
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.77821401024839e-06,
+      "loss": 0.102,
+      "step": 7903
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.775330931980112e-06,
+      "loss": 0.0724,
+      "step": 7904
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.772449197931031e-06,
+      "loss": 0.1096,
+      "step": 7905
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7695688084180134e-06,
+      "loss": 0.0523,
+      "step": 7906
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.766689763757788e-06,
+      "loss": 0.0662,
+      "step": 7907
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.763812064266938e-06,
+      "loss": 0.0687,
+      "step": 7908
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7609357102618893e-06,
+      "loss": 0.0424,
+      "step": 7909
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.758060702058922e-06,
+      "loss": 0.0819,
+      "step": 7910
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7551870399741734e-06,
+      "loss": 0.1227,
+      "step": 7911
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7523147243236326e-06,
+      "loss": 0.0527,
+      "step": 7912
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7494437554231377e-06,
+      "loss": 0.0934,
+      "step": 7913
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.746574133588379e-06,
+      "loss": 0.0882,
+      "step": 7914
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7437058591348945e-06,
+      "loss": 0.1301,
+      "step": 7915
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7408389323780798e-06,
+      "loss": 0.0538,
+      "step": 7916
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.737973353633187e-06,
+      "loss": 0.1052,
+      "step": 7917
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7351091232153056e-06,
+      "loss": 0.1345,
+      "step": 7918
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7322462414393897e-06,
+      "loss": 0.0901,
+      "step": 7919
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.729384708620243e-06,
+      "loss": 0.095,
+      "step": 7920
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.726524525072514e-06,
+      "loss": 0.1417,
+      "step": 7921
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.723665691110705e-06,
+      "loss": 0.1112,
+      "step": 7922
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.720808207049176e-06,
+      "loss": 0.1078,
+      "step": 7923
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7179520732021353e-06,
+      "loss": 0.0737,
+      "step": 7924
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7150972898836433e-06,
+      "loss": 0.0819,
+      "step": 7925
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.712243857407608e-06,
+      "loss": 0.1182,
+      "step": 7926
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.709391776087788e-06,
+      "loss": 0.052,
+      "step": 7927
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7065410462378022e-06,
+      "loss": 0.0726,
+      "step": 7928
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7036916681711133e-06,
+      "loss": 0.0786,
+      "step": 7929
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7008436422010426e-06,
+      "loss": 0.1201,
+      "step": 7930
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.697996968640754e-06,
+      "loss": 0.1181,
+      "step": 7931
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.6951516478032613e-06,
+      "loss": 0.0918,
+      "step": 7932
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.6923076800014403e-06,
+      "loss": 0.1136,
+      "step": 7933
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.6894650655480155e-06,
+      "loss": 0.0987,
+      "step": 7934
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.6866238047555515e-06,
+      "loss": 0.0743,
+      "step": 7935
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.6837838979364804e-06,
+      "loss": 0.0587,
+      "step": 7936
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6809453454030696e-06,
+      "loss": 0.0831,
+      "step": 7937
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6781081474674528e-06,
+      "loss": 0.0804,
+      "step": 7938
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.675272304441599e-06,
+      "loss": 0.0532,
+      "step": 7939
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.672437816637338e-06,
+      "loss": 0.0534,
+      "step": 7940
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.669604684366353e-06,
+      "loss": 0.0905,
+      "step": 7941
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.666772907940174e-06,
+      "loss": 0.0836,
+      "step": 7942
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.663942487670181e-06,
+      "loss": 0.0805,
+      "step": 7943
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6611134238676e-06,
+      "loss": 0.0721,
+      "step": 7944
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.658285716843518e-06,
+      "loss": 0.1163,
+      "step": 7945
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6554593669088723e-06,
+      "loss": 0.0878,
+      "step": 7946
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6526343743744387e-06,
+      "loss": 0.0863,
+      "step": 7947
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.649810739550862e-06,
+      "loss": 0.0685,
+      "step": 7948
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6469884627486176e-06,
+      "loss": 0.1119,
+      "step": 7949
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.64416754427805e-06,
+      "loss": 0.081,
+      "step": 7950
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.64134798444934e-06,
+      "loss": 0.0567,
+      "step": 7951
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6385297835725273e-06,
+      "loss": 0.083,
+      "step": 7952
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.635712941957505e-06,
+      "loss": 0.0772,
+      "step": 7953
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.632897459914008e-06,
+      "loss": 0.1313,
+      "step": 7954
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.63008333775162e-06,
+      "loss": 0.0678,
+      "step": 7955
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6272705757797856e-06,
+      "loss": 0.0712,
+      "step": 7956
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6244591743077956e-06,
+      "loss": 0.0521,
+      "step": 7957
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6216491336447927e-06,
+      "loss": 0.1318,
+      "step": 7958
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.618840454099765e-06,
+      "loss": 0.0941,
+      "step": 7959
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6160331359815503e-06,
+      "loss": 0.1027,
+      "step": 7960
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6132271795988415e-06,
+      "loss": 0.0747,
+      "step": 7961
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6104225852601836e-06,
+      "loss": 0.1313,
+      "step": 7962
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.60761935327397e-06,
+      "loss": 0.1108,
+      "step": 7963
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.604817483948439e-06,
+      "loss": 0.0446,
+      "step": 7964
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6020169775916808e-06,
+      "loss": 0.1118,
+      "step": 7965
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5992178345116407e-06,
+      "loss": 0.166,
+      "step": 7966
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5964200550161142e-06,
+      "loss": 0.0977,
+      "step": 7967
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.593623639412737e-06,
+      "loss": 0.0825,
+      "step": 7968
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5908285880090067e-06,
+      "loss": 0.1162,
+      "step": 7969
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.588034901112268e-06,
+      "loss": 0.1332,
+      "step": 7970
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5852425790297103e-06,
+      "loss": 0.1182,
+      "step": 7971
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5824516220683715e-06,
+      "loss": 0.0715,
+      "step": 7972
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5796620305351497e-06,
+      "loss": 0.0553,
+      "step": 7973
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.576873804736785e-06,
+      "loss": 0.0836,
+      "step": 7974
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5740869449798754e-06,
+      "loss": 0.0612,
+      "step": 7975
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.571301451570857e-06,
+      "loss": 0.0981,
+      "step": 7976
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5685173248160186e-06,
+      "loss": 0.0529,
+      "step": 7977
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5657345650215068e-06,
+      "loss": 0.0284,
+      "step": 7978
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5629531724933135e-06,
+      "loss": 0.1006,
+      "step": 7979
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5601731475372723e-06,
+      "loss": 0.0508,
+      "step": 7980
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5573944904590842e-06,
+      "loss": 0.067,
+      "step": 7981
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.554617201564277e-06,
+      "loss": 0.0772,
+      "step": 7982
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5518412811582507e-06,
+      "loss": 0.1075,
+      "step": 7983
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5490667295462378e-06,
+      "loss": 0.0538,
+      "step": 7984
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.546293547033327e-06,
+      "loss": 0.127,
+      "step": 7985
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5435217339244613e-06,
+      "loss": 0.1276,
+      "step": 7986
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5407512905244256e-06,
+      "loss": 0.1286,
+      "step": 7987
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5379822171378524e-06,
+      "loss": 0.0672,
+      "step": 7988
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.53521451406923e-06,
+      "loss": 0.0736,
+      "step": 7989
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.532448181622897e-06,
+      "loss": 0.0609,
+      "step": 7990
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.529683220103034e-06,
+      "loss": 0.0546,
+      "step": 7991
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5269196298136855e-06,
+      "loss": 0.0654,
+      "step": 7992
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5241574110587177e-06,
+      "loss": 0.0679,
+      "step": 7993
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5213965641418716e-06,
+      "loss": 0.0583,
+      "step": 7994
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5186370893667294e-06,
+      "loss": 0.0868,
+      "step": 7995
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5158789870367227e-06,
+      "loss": 0.1082,
+      "step": 7996
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5131222574551276e-06,
+      "loss": 0.0628,
+      "step": 7997
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.510366900925077e-06,
+      "loss": 0.1077,
+      "step": 7998
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5076129177495425e-06,
+      "loss": 0.0921,
+      "step": 7999
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.50486030823136e-06,
+      "loss": 0.113,
+      "step": 8000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5021090726731943e-06,
+      "loss": 0.1248,
+      "step": 8001
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4993592113775778e-06,
+      "loss": 0.1117,
+      "step": 8002
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4966107246468828e-06,
+      "loss": 0.0747,
+      "step": 8003
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.493863612783333e-06,
+      "loss": 0.1221,
+      "step": 8004
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.491117876088992e-06,
+      "loss": 0.0999,
+      "step": 8005
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.488373514865786e-06,
+      "loss": 0.1133,
+      "step": 8006
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4856305294154824e-06,
+      "loss": 0.0824,
+      "step": 8007
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.482888920039703e-06,
+      "loss": 0.0914,
+      "step": 8008
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4801486870399104e-06,
+      "loss": 0.0878,
+      "step": 8009
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.477409830717415e-06,
+      "loss": 0.0728,
+      "step": 8010
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4746723513733856e-06,
+      "loss": 0.1438,
+      "step": 8011
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.471936249308835e-06,
+      "loss": 0.081,
+      "step": 8012
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4692015248246175e-06,
+      "loss": 0.1109,
+      "step": 8013
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.466468178221451e-06,
+      "loss": 0.1537,
+      "step": 8014
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.463736209799885e-06,
+      "loss": 0.0691,
+      "step": 8015
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4610056198603326e-06,
+      "loss": 0.0656,
+      "step": 8016
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.458276408703042e-06,
+      "loss": 0.0895,
+      "step": 8017
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.455548576628116e-06,
+      "loss": 0.0987,
+      "step": 8018
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.45282212393551e-06,
+      "loss": 0.0812,
+      "step": 8019
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4500970509250266e-06,
+      "loss": 0.1008,
+      "step": 8020
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.447373357896308e-06,
+      "loss": 0.1202,
+      "step": 8021
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4446510451488472e-06,
+      "loss": 0.119,
+      "step": 8022
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.441930112981993e-06,
+      "loss": 0.1002,
+      "step": 8023
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4392105616949397e-06,
+      "loss": 0.136,
+      "step": 8024
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.436492391586727e-06,
+      "loss": 0.0863,
+      "step": 8025
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4337756029562436e-06,
+      "loss": 0.0539,
+      "step": 8026
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.43106019610222e-06,
+      "loss": 0.0981,
+      "step": 8027
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4283461713232472e-06,
+      "loss": 0.128,
+      "step": 8028
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.425633528917762e-06,
+      "loss": 0.1171,
+      "step": 8029
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4229222691840356e-06,
+      "loss": 0.0982,
+      "step": 8030
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4202123924202065e-06,
+      "loss": 0.0395,
+      "step": 8031
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.417503898924244e-06,
+      "loss": 0.0657,
+      "step": 8032
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4147967889939783e-06,
+      "loss": 0.1352,
+      "step": 8033
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4120910629270764e-06,
+      "loss": 0.0837,
+      "step": 8034
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.4093867210210625e-06,
+      "loss": 0.1066,
+      "step": 8035
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.406683763573308e-06,
+      "loss": 0.1105,
+      "step": 8036
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.4039821908810237e-06,
+      "loss": 0.0763,
+      "step": 8037
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.4012820032412725e-06,
+      "loss": 0.0798,
+      "step": 8038
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.398583200950968e-06,
+      "loss": 0.0433,
+      "step": 8039
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3958857843068693e-06,
+      "loss": 0.1094,
+      "step": 8040
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.393189753605588e-06,
+      "loss": 0.1292,
+      "step": 8041
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.390495109143572e-06,
+      "loss": 0.0674,
+      "step": 8042
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.387801851217122e-06,
+      "loss": 0.0834,
+      "step": 8043
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3851099801223912e-06,
+      "loss": 0.0546,
+      "step": 8044
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3824194961553787e-06,
+      "loss": 0.0653,
+      "step": 8045
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3797303996119217e-06,
+      "loss": 0.0911,
+      "step": 8046
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.377042690787718e-06,
+      "loss": 0.0771,
+      "step": 8047
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.374356369978307e-06,
+      "loss": 0.058,
+      "step": 8048
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3716714374790714e-06,
+      "loss": 0.0801,
+      "step": 8049
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.36898789358525e-06,
+      "loss": 0.0457,
+      "step": 8050
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.366305738591918e-06,
+      "loss": 0.0864,
+      "step": 8051
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.363624972794008e-06,
+      "loss": 0.0695,
+      "step": 8052
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3609455964862993e-06,
+      "loss": 0.0704,
+      "step": 8053
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3582676099634094e-06,
+      "loss": 0.106,
+      "step": 8054
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.355591013519806e-06,
+      "loss": 0.07,
+      "step": 8055
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3529158074498124e-06,
+      "loss": 0.1061,
+      "step": 8056
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.35024199204759e-06,
+      "loss": 0.0766,
+      "step": 8057
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.347569567607156e-06,
+      "loss": 0.078,
+      "step": 8058
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3448985344223634e-06,
+      "loss": 0.092,
+      "step": 8059
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.342228892786916e-06,
+      "loss": 0.0529,
+      "step": 8060
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.339560642994368e-06,
+      "loss": 0.0645,
+      "step": 8061
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.336893785338125e-06,
+      "loss": 0.0505,
+      "step": 8062
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.334228320111425e-06,
+      "loss": 0.1006,
+      "step": 8063
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.331564247607369e-06,
+      "loss": 0.0921,
+      "step": 8064
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.328901568118891e-06,
+      "loss": 0.0959,
+      "step": 8065
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.326240281938781e-06,
+      "loss": 0.0448,
+      "step": 8066
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3235803893596714e-06,
+      "loss": 0.1028,
+      "step": 8067
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3209218906740422e-06,
+      "loss": 0.0962,
+      "step": 8068
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.318264786174223e-06,
+      "loss": 0.0562,
+      "step": 8069
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3156090761523933e-06,
+      "loss": 0.0712,
+      "step": 8070
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.312954760900562e-06,
+      "loss": 0.1013,
+      "step": 8071
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3103018407106014e-06,
+      "loss": 0.1158,
+      "step": 8072
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.307650315874225e-06,
+      "loss": 0.1003,
+      "step": 8073
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3050001866829982e-06,
+      "loss": 0.0834,
+      "step": 8074
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3023514534283214e-06,
+      "loss": 0.0709,
+      "step": 8075
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.299704116401454e-06,
+      "loss": 0.0349,
+      "step": 8076
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2970581758934893e-06,
+      "loss": 0.0674,
+      "step": 8077
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2944136321953808e-06,
+      "loss": 0.13,
+      "step": 8078
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.291770485597914e-06,
+      "loss": 0.0747,
+      "step": 8079
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.289128736391733e-06,
+      "loss": 0.0891,
+      "step": 8080
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.286488384867324e-06,
+      "loss": 0.0731,
+      "step": 8081
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.283849431315015e-06,
+      "loss": 0.0868,
+      "step": 8082
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.28121187602499e-06,
+      "loss": 0.0854,
+      "step": 8083
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2785757192872676e-06,
+      "loss": 0.0892,
+      "step": 8084
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2759409613917188e-06,
+      "loss": 0.1238,
+      "step": 8085
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.273307602628067e-06,
+      "loss": 0.1081,
+      "step": 8086
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.270675643285869e-06,
+      "loss": 0.0842,
+      "step": 8087
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.268045083654533e-06,
+      "loss": 0.0626,
+      "step": 8088
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2654159240233164e-06,
+      "loss": 0.0554,
+      "step": 8089
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.262788164681319e-06,
+      "loss": 0.0472,
+      "step": 8090
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2601618059174957e-06,
+      "loss": 0.182,
+      "step": 8091
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.257536848020631e-06,
+      "loss": 0.0909,
+      "step": 8092
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2549132912793644e-06,
+      "loss": 0.0428,
+      "step": 8093
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.252291135982184e-06,
+      "loss": 0.0453,
+      "step": 8094
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2496703824174235e-06,
+      "loss": 0.1051,
+      "step": 8095
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.247051030873253e-06,
+      "loss": 0.1183,
+      "step": 8096
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2444330816376977e-06,
+      "loss": 0.0718,
+      "step": 8097
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.241816534998632e-06,
+      "loss": 0.0455,
+      "step": 8098
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.239201391243765e-06,
+      "loss": 0.0838,
+      "step": 8099
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.236587650660653e-06,
+      "loss": 0.1352,
+      "step": 8100
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2339753135367053e-06,
+      "loss": 0.0682,
+      "step": 8101
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2313643801591747e-06,
+      "loss": 0.0548,
+      "step": 8102
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2287548508151604e-06,
+      "loss": 0.0782,
+      "step": 8103
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2261467257916045e-06,
+      "loss": 0.1195,
+      "step": 8104
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2235400053752874e-06,
+      "loss": 0.0648,
+      "step": 8105
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2209346898528484e-06,
+      "loss": 0.1563,
+      "step": 8106
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.218330779510771e-06,
+      "loss": 0.0932,
+      "step": 8107
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2157282746353735e-06,
+      "loss": 0.0857,
+      "step": 8108
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.21312717551283e-06,
+      "loss": 0.0845,
+      "step": 8109
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.210527482429152e-06,
+      "loss": 0.0358,
+      "step": 8110
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.207929195670204e-06,
+      "loss": 0.159,
+      "step": 8111
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.205332315521696e-06,
+      "loss": 0.1213,
+      "step": 8112
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2027368422691707e-06,
+      "loss": 0.0993,
+      "step": 8113
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.200142776198035e-06,
+      "loss": 0.0442,
+      "step": 8114
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1975501175935216e-06,
+      "loss": 0.1193,
+      "step": 8115
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1949588667407276e-06,
+      "loss": 0.0489,
+      "step": 8116
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.192369023924576e-06,
+      "loss": 0.0791,
+      "step": 8117
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.189780589429851e-06,
+      "loss": 0.091,
+      "step": 8118
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.187193563541176e-06,
+      "loss": 0.0696,
+      "step": 8119
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.184607946543022e-06,
+      "loss": 0.0525,
+      "step": 8120
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.182023738719693e-06,
+      "loss": 0.054,
+      "step": 8121
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1794409403553516e-06,
+      "loss": 0.1013,
+      "step": 8122
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.176859551734003e-06,
+      "loss": 0.0941,
+      "step": 8123
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.174279573139498e-06,
+      "loss": 0.1179,
+      "step": 8124
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1717010048555208e-06,
+      "loss": 0.1292,
+      "step": 8125
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1691238471656207e-06,
+      "loss": 0.0681,
+      "step": 8126
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1665481003531716e-06,
+      "loss": 0.08,
+      "step": 8127
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1639737647014094e-06,
+      "loss": 0.0782,
+      "step": 8128
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1614008404933995e-06,
+      "loss": 0.0611,
+      "step": 8129
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1588293280120614e-06,
+      "loss": 0.1366,
+      "step": 8130
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1562592275401627e-06,
+      "loss": 0.0838,
+      "step": 8131
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1536905393603074e-06,
+      "loss": 0.0439,
+      "step": 8132
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.151123263754943e-06,
+      "loss": 0.0963,
+      "step": 8133
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.148557401006368e-06,
+      "loss": 0.0855,
+      "step": 8134
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1459929513967276e-06,
+      "loss": 0.0836,
+      "step": 8135
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1434299152080074e-06,
+      "loss": 0.1145,
+      "step": 8136
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1408682927220336e-06,
+      "loss": 0.0908,
+      "step": 8137
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.138308084220481e-06,
+      "loss": 0.0994,
+      "step": 8138
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.135749289984871e-06,
+      "loss": 0.0612,
+      "step": 8139
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1331919102965673e-06,
+      "loss": 0.0716,
+      "step": 8140
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.130635945436783e-06,
+      "loss": 0.0693,
+      "step": 8141
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1280813956865646e-06,
+      "loss": 0.0553,
+      "step": 8142
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.125528261326808e-06,
+      "loss": 0.1062,
+      "step": 8143
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1229765426382563e-06,
+      "loss": 0.098,
+      "step": 8144
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1204262399015006e-06,
+      "loss": 0.0963,
+      "step": 8145
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1178773533969633e-06,
+      "loss": 0.057,
+      "step": 8146
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.115329883404923e-06,
+      "loss": 0.109,
+      "step": 8147
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1127838302055024e-06,
+      "loss": 0.0852,
+      "step": 8148
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1102391940786582e-06,
+      "loss": 0.0427,
+      "step": 8149
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1076959753041976e-06,
+      "loss": 0.0736,
+      "step": 8150
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.105154174161772e-06,
+      "loss": 0.0851,
+      "step": 8151
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.102613790930878e-06,
+      "loss": 0.0927,
+      "step": 8152
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1000748258908585e-06,
+      "loss": 0.0519,
+      "step": 8153
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0975372793208943e-06,
+      "loss": 0.0491,
+      "step": 8154
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0950011515000077e-06,
+      "loss": 0.0604,
+      "step": 8155
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.092466442707075e-06,
+      "loss": 0.0976,
+      "step": 8156
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.089933153220816e-06,
+      "loss": 0.1009,
+      "step": 8157
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0874012833197814e-06,
+      "loss": 0.1174,
+      "step": 8158
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.084870833282382e-06,
+      "loss": 0.1141,
+      "step": 8159
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.082341803386856e-06,
+      "loss": 0.1109,
+      "step": 8160
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.079814193911306e-06,
+      "loss": 0.1099,
+      "step": 8161
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0772880051336563e-06,
+      "loss": 0.103,
+      "step": 8162
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0747632373316904e-06,
+      "loss": 0.1244,
+      "step": 8163
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.072239890783033e-06,
+      "loss": 0.0662,
+      "step": 8164
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.069717965765147e-06,
+      "loss": 0.064,
+      "step": 8165
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.067197462555339e-06,
+      "loss": 0.0852,
+      "step": 8166
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0646783814307672e-06,
+      "loss": 0.079,
+      "step": 8167
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0621607226684254e-06,
+      "loss": 0.1059,
+      "step": 8168
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0596444865451614e-06,
+      "loss": 0.1192,
+      "step": 8169
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.057129673337653e-06,
+      "loss": 0.1027,
+      "step": 8170
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.054616283322426e-06,
+      "loss": 0.0671,
+      "step": 8171
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0521043167758544e-06,
+      "loss": 0.0972,
+      "step": 8172
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.049593773974155e-06,
+      "loss": 0.1118,
+      "step": 8173
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0470846551933863e-06,
+      "loss": 0.0784,
+      "step": 8174
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.044576960709445e-06,
+      "loss": 0.0359,
+      "step": 8175
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0420706907980817e-06,
+      "loss": 0.0893,
+      "step": 8176
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.039565845734878e-06,
+      "loss": 0.0518,
+      "step": 8177
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0370624257952753e-06,
+      "loss": 0.0952,
+      "step": 8178
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0345604312545377e-06,
+      "loss": 0.1224,
+      "step": 8179
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0320598623877876e-06,
+      "loss": 0.0637,
+      "step": 8180
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0295607194699935e-06,
+      "loss": 0.0669,
+      "step": 8181
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.027063002775951e-06,
+      "loss": 0.096,
+      "step": 8182
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0245667125803086e-06,
+      "loss": 0.1033,
+      "step": 8183
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0220718491575608e-06,
+      "loss": 0.0767,
+      "step": 8184
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.019578412782038e-06,
+      "loss": 0.0839,
+      "step": 8185
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0170864037279236e-06,
+      "loss": 0.0671,
+      "step": 8186
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.014595822269236e-06,
+      "loss": 0.1366,
+      "step": 8187
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0121066686798306e-06,
+      "loss": 0.1262,
+      "step": 8188
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0096189432334194e-06,
+      "loss": 0.1072,
+      "step": 8189
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0071326462035567e-06,
+      "loss": 0.0641,
+      "step": 8190
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0046477778636253e-06,
+      "loss": 0.102,
+      "step": 8191
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0021643384868666e-06,
+      "loss": 0.0589,
+      "step": 8192
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.999682328346354e-06,
+      "loss": 0.0913,
+      "step": 8193
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9972017477150138e-06,
+      "loss": 0.0587,
+      "step": 8194
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9947225968656037e-06,
+      "loss": 0.044,
+      "step": 8195
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9922448760707314e-06,
+      "loss": 0.0437,
+      "step": 8196
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9897685856028476e-06,
+      "loss": 0.0867,
+      "step": 8197
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9872937257342493e-06,
+      "loss": 0.0622,
+      "step": 8198
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.984820296737061e-06,
+      "loss": 0.0656,
+      "step": 8199
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9823482988832615e-06,
+      "loss": 0.1074,
+      "step": 8200
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.979877732444675e-06,
+      "loss": 0.0629,
+      "step": 8201
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9774085976929624e-06,
+      "loss": 0.0805,
+      "step": 8202
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.974940894899632e-06,
+      "loss": 0.1273,
+      "step": 8203
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.972474624336026e-06,
+      "loss": 0.0489,
+      "step": 8204
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9700097862733345e-06,
+      "loss": 0.0945,
+      "step": 8205
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.967546380982591e-06,
+      "loss": 0.0887,
+      "step": 8206
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9650844087346755e-06,
+      "loss": 0.0534,
+      "step": 8207
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.962623869800296e-06,
+      "loss": 0.0778,
+      "step": 8208
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9601647644500226e-06,
+      "loss": 0.0949,
+      "step": 8209
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9577070929542464e-06,
+      "loss": 0.0419,
+      "step": 8210
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9552508555832223e-06,
+      "loss": 0.0513,
+      "step": 8211
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9527960526070294e-06,
+      "loss": 0.1219,
+      "step": 8212
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9503426842956e-06,
+      "loss": 0.0861,
+      "step": 8213
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9478907509187073e-06,
+      "loss": 0.0407,
+      "step": 8214
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9454402527459632e-06,
+      "loss": 0.1057,
+      "step": 8215
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9429911900468188e-06,
+      "loss": 0.0396,
+      "step": 8216
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9405435630905775e-06,
+      "loss": 0.0538,
+      "step": 8217
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9380973721463757e-06,
+      "loss": 0.0971,
+      "step": 8218
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.935652617483203e-06,
+      "loss": 0.0665,
+      "step": 8219
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.933209299369876e-06,
+      "loss": 0.0962,
+      "step": 8220
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.930767418075061e-06,
+      "loss": 0.1107,
+      "step": 8221
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9283269738672674e-06,
+      "loss": 0.0973,
+      "step": 8222
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9258879670148503e-06,
+      "loss": 0.0988,
+      "step": 8223
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9234503977859935e-06,
+      "loss": 0.0452,
+      "step": 8224
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.921014266448736e-06,
+      "loss": 0.0703,
+      "step": 8225
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9185795732709545e-06,
+      "loss": 0.0603,
+      "step": 8226
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9161463185203652e-06,
+      "loss": 0.1079,
+      "step": 8227
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9137145024645263e-06,
+      "loss": 0.0478,
+      "step": 8228
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9112841253708384e-06,
+      "loss": 0.0533,
+      "step": 8229
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.908855187506548e-06,
+      "loss": 0.0944,
+      "step": 8230
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9064276891387428e-06,
+      "loss": 0.1023,
+      "step": 8231
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9040016305343445e-06,
+      "loss": 0.1071,
+      "step": 8232
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9015770119601195e-06,
+      "loss": 0.0582,
+      "step": 8233
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8991538336826796e-06,
+      "loss": 0.0852,
+      "step": 8234
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8967320959684804e-06,
+      "loss": 0.0622,
+      "step": 8235
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8943117990838132e-06,
+      "loss": 0.0966,
+      "step": 8236
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8918929432948123e-06,
+      "loss": 0.0772,
+      "step": 8237
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8894755288674499e-06,
+      "loss": 0.138,
+      "step": 8238
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.887059556067548e-06,
+      "loss": 0.0965,
+      "step": 8239
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8846450251607678e-06,
+      "loss": 0.0769,
+      "step": 8240
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.882231936412605e-06,
+      "loss": 0.1045,
+      "step": 8241
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8798202900884077e-06,
+      "loss": 0.0379,
+      "step": 8242
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8774100864533534e-06,
+      "loss": 0.0948,
+      "step": 8243
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.875001325772472e-06,
+      "loss": 0.0705,
+      "step": 8244
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8725940083106247e-06,
+      "loss": 0.122,
+      "step": 8245
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8701881343325217e-06,
+      "loss": 0.0792,
+      "step": 8246
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8677837041027145e-06,
+      "loss": 0.0849,
+      "step": 8247
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.865380717885592e-06,
+      "loss": 0.0867,
+      "step": 8248
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8629791759453806e-06,
+      "loss": 0.0691,
+      "step": 8249
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8605790785461562e-06,
+      "loss": 0.0872,
+      "step": 8250
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8581804259518326e-06,
+      "loss": 0.0692,
+      "step": 8251
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8557832184261685e-06,
+      "loss": 0.0482,
+      "step": 8252
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8533874562327519e-06,
+      "loss": 0.0712,
+      "step": 8253
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8509931396350267e-06,
+      "loss": 0.1388,
+      "step": 8254
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8486002688962656e-06,
+      "loss": 0.1308,
+      "step": 8255
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8462088442795915e-06,
+      "loss": 0.0965,
+      "step": 8256
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.843818866047961e-06,
+      "loss": 0.0654,
+      "step": 8257
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.841430334464177e-06,
+      "loss": 0.079,
+      "step": 8258
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8390432497908844e-06,
+      "loss": 0.1144,
+      "step": 8259
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8366576122905582e-06,
+      "loss": 0.0592,
+      "step": 8260
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8342734222255303e-06,
+      "loss": 0.0776,
+      "step": 8261
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8318906798579594e-06,
+      "loss": 0.0612,
+      "step": 8262
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8295093854498523e-06,
+      "loss": 0.1073,
+      "step": 8263
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.82712953926306e-06,
+      "loss": 0.0714,
+      "step": 8264
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8247511415592628e-06,
+      "loss": 0.103,
+      "step": 8265
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8223741925999883e-06,
+      "loss": 0.0978,
+      "step": 8266
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8199986926466072e-06,
+      "loss": 0.1297,
+      "step": 8267
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8176246419603292e-06,
+      "loss": 0.0822,
+      "step": 8268
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8152520408022038e-06,
+      "loss": 0.0781,
+      "step": 8269
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8128808894331223e-06,
+      "loss": 0.0937,
+      "step": 8270
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8105111881138093e-06,
+      "loss": 0.0812,
+      "step": 8271
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8081429371048397e-06,
+      "loss": 0.0705,
+      "step": 8272
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8057761366666287e-06,
+      "loss": 0.0824,
+      "step": 8273
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.803410787059423e-06,
+      "loss": 0.0898,
+      "step": 8274
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8010468885433179e-06,
+      "loss": 0.062,
+      "step": 8275
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.7986844413782504e-06,
+      "loss": 0.1182,
+      "step": 8276
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.7963234458239914e-06,
+      "loss": 0.0742,
+      "step": 8277
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.7939639021401495e-06,
+      "loss": 0.0513,
+      "step": 8278
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7916058105861837e-06,
+      "loss": 0.0894,
+      "step": 8279
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7892491714213903e-06,
+      "loss": 0.0689,
+      "step": 8280
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7868939849049054e-06,
+      "loss": 0.1068,
+      "step": 8281
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7845402512956999e-06,
+      "loss": 0.138,
+      "step": 8282
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7821879708525885e-06,
+      "loss": 0.037,
+      "step": 8283
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.779837143834231e-06,
+      "loss": 0.0708,
+      "step": 8284
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7774877704991222e-06,
+      "loss": 0.1018,
+      "step": 8285
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7751398511055973e-06,
+      "loss": 0.0732,
+      "step": 8286
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7727933859118344e-06,
+      "loss": 0.0965,
+      "step": 8287
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7704483751758443e-06,
+      "loss": 0.0729,
+      "step": 8288
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7681048191554905e-06,
+      "loss": 0.0463,
+      "step": 8289
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.765762718108465e-06,
+      "loss": 0.0756,
+      "step": 8290
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7634220722923038e-06,
+      "loss": 0.0928,
+      "step": 8291
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7610828819643877e-06,
+      "loss": 0.0627,
+      "step": 8292
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7587451473819271e-06,
+      "loss": 0.101,
+      "step": 8293
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7564088688019852e-06,
+      "loss": 0.0657,
+      "step": 8294
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7540740464814497e-06,
+      "loss": 0.0515,
+      "step": 8295
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7517406806770619e-06,
+      "loss": 0.0653,
+      "step": 8296
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7494087716454e-06,
+      "loss": 0.0701,
+      "step": 8297
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7470783196428768e-06,
+      "loss": 0.0908,
+      "step": 8298
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7447493249257445e-06,
+      "loss": 0.0581,
+      "step": 8299
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7424217877500997e-06,
+      "loss": 0.0783,
+      "step": 8300
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.740095708371881e-06,
+      "loss": 0.0725,
+      "step": 8301
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7377710870468637e-06,
+      "loss": 0.0584,
+      "step": 8302
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.735447924030657e-06,
+      "loss": 0.0667,
+      "step": 8303
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7331262195787195e-06,
+      "loss": 0.0632,
+      "step": 8304
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7308059739463405e-06,
+      "loss": 0.063,
+      "step": 8305
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7284871873886577e-06,
+      "loss": 0.1201,
+      "step": 8306
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7261698601606386e-06,
+      "loss": 0.1401,
+      "step": 8307
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7238539925170976e-06,
+      "loss": 0.1043,
+      "step": 8308
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7215395847126914e-06,
+      "loss": 0.0716,
+      "step": 8309
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7192266370019077e-06,
+      "loss": 0.09,
+      "step": 8310
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7169151496390735e-06,
+      "loss": 0.104,
+      "step": 8311
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7146051228783598e-06,
+      "loss": 0.0538,
+      "step": 8312
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.712296556973781e-06,
+      "loss": 0.1027,
+      "step": 8313
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7099894521791847e-06,
+      "loss": 0.0677,
+      "step": 8314
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7076838087482588e-06,
+      "loss": 0.0649,
+      "step": 8315
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7053796269345268e-06,
+      "loss": 0.0669,
+      "step": 8316
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7030769069913578e-06,
+      "loss": 0.0623,
+      "step": 8317
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7007756491719623e-06,
+      "loss": 0.084,
+      "step": 8318
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6984758537293783e-06,
+      "loss": 0.1265,
+      "step": 8319
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.696177520916498e-06,
+      "loss": 0.0829,
+      "step": 8320
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6938806509860367e-06,
+      "loss": 0.0828,
+      "step": 8321
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6915852441905615e-06,
+      "loss": 0.0869,
+      "step": 8322
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6892913007824779e-06,
+      "loss": 0.0695,
+      "step": 8323
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6869988210140203e-06,
+      "loss": 0.0507,
+      "step": 8324
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.684707805137274e-06,
+      "loss": 0.0759,
+      "step": 8325
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6824182534041537e-06,
+      "loss": 0.1032,
+      "step": 8326
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.680130166066422e-06,
+      "loss": 0.0976,
+      "step": 8327
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6778435433756707e-06,
+      "loss": 0.0919,
+      "step": 8328
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6755583855833385e-06,
+      "loss": 0.0628,
+      "step": 8329
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6732746929406994e-06,
+      "loss": 0.0694,
+      "step": 8330
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.670992465698873e-06,
+      "loss": 0.1099,
+      "step": 8331
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6687117041088062e-06,
+      "loss": 0.0986,
+      "step": 8332
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.666432408421289e-06,
+      "loss": 0.1536,
+      "step": 8333
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6641545788869556e-06,
+      "loss": 0.0436,
+      "step": 8334
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6618782157562757e-06,
+      "loss": 0.0932,
+      "step": 8335
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6596033192795522e-06,
+      "loss": 0.0884,
+      "step": 8336
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6573298897069389e-06,
+      "loss": 0.0909,
+      "step": 8337
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6550579272884137e-06,
+      "loss": 0.0713,
+      "step": 8338
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6527874322738068e-06,
+      "loss": 0.0855,
+      "step": 8339
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6505184049127754e-06,
+      "loss": 0.0633,
+      "step": 8340
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6482508454548229e-06,
+      "loss": 0.1087,
+      "step": 8341
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6459847541492933e-06,
+      "loss": 0.0785,
+      "step": 8342
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.643720131245361e-06,
+      "loss": 0.0878,
+      "step": 8343
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6414569769920413e-06,
+      "loss": 0.097,
+      "step": 8344
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6391952916381903e-06,
+      "loss": 0.0823,
+      "step": 8345
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6369350754325025e-06,
+      "loss": 0.0924,
+      "step": 8346
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6346763286235157e-06,
+      "loss": 0.1293,
+      "step": 8347
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6324190514595944e-06,
+      "loss": 0.1168,
+      "step": 8348
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.630163244188947e-06,
+      "loss": 0.093,
+      "step": 8349
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.627908907059625e-06,
+      "loss": 0.0839,
+      "step": 8350
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6256560403195097e-06,
+      "loss": 0.0702,
+      "step": 8351
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6234046442163336e-06,
+      "loss": 0.053,
+      "step": 8352
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.62115471899765e-06,
+      "loss": 0.0902,
+      "step": 8353
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6189062649108676e-06,
+      "loss": 0.1392,
+      "step": 8354
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6166592822032172e-06,
+      "loss": 0.0682,
+      "step": 8355
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6144137711217843e-06,
+      "loss": 0.113,
+      "step": 8356
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6121697319134782e-06,
+      "loss": 0.1039,
+      "step": 8357
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.609927164825053e-06,
+      "loss": 0.0502,
+      "step": 8358
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6076860701031048e-06,
+      "loss": 0.0966,
+      "step": 8359
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6054464479940617e-06,
+      "loss": 0.0802,
+      "step": 8360
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6032082987441864e-06,
+      "loss": 0.0339,
+      "step": 8361
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.600971622599589e-06,
+      "loss": 0.0411,
+      "step": 8362
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5987364198062143e-06,
+      "loss": 0.0914,
+      "step": 8363
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.596502690609844e-06,
+      "loss": 0.0706,
+      "step": 8364
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5942704352561e-06,
+      "loss": 0.1418,
+      "step": 8365
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5920396539904326e-06,
+      "loss": 0.0615,
+      "step": 8366
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5898103470581405e-06,
+      "loss": 0.0729,
+      "step": 8367
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5875825147043644e-06,
+      "loss": 0.1014,
+      "step": 8368
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5853561571740665e-06,
+      "loss": 0.0548,
+      "step": 8369
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.583131274712063e-06,
+      "loss": 0.0991,
+      "step": 8370
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5809078675629945e-06,
+      "loss": 0.0733,
+      "step": 8371
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5786859359713523e-06,
+      "loss": 0.1227,
+      "step": 8372
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5764654801814527e-06,
+      "loss": 0.0666,
+      "step": 8373
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5742465004374584e-06,
+      "loss": 0.0551,
+      "step": 8374
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.572028996983373e-06,
+      "loss": 0.1223,
+      "step": 8375
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5698129700630275e-06,
+      "loss": 0.0268,
+      "step": 8376
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.567598419920091e-06,
+      "loss": 0.0692,
+      "step": 8377
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.56538534679808e-06,
+      "loss": 0.066,
+      "step": 8378
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.56317375094034e-06,
+      "loss": 0.0607,
+      "step": 8379
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5609636325900628e-06,
+      "loss": 0.0535,
+      "step": 8380
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5587549919902661e-06,
+      "loss": 0.0865,
+      "step": 8381
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5565478293838137e-06,
+      "loss": 0.1139,
+      "step": 8382
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5543421450134032e-06,
+      "loss": 0.0828,
+      "step": 8383
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.552137939121569e-06,
+      "loss": 0.0426,
+      "step": 8384
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5499352119506904e-06,
+      "loss": 0.0902,
+      "step": 8385
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5477339637429706e-06,
+      "loss": 0.0766,
+      "step": 8386
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5455341947404656e-06,
+      "loss": 0.0468,
+      "step": 8387
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5433359051850553e-06,
+      "loss": 0.0985,
+      "step": 8388
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5411390953184651e-06,
+      "loss": 0.0965,
+      "step": 8389
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.538943765382253e-06,
+      "loss": 0.0628,
+      "step": 8390
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5367499156178194e-06,
+      "loss": 0.0391,
+      "step": 8391
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5345575462663997e-06,
+      "loss": 0.0347,
+      "step": 8392
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5323666575690626e-06,
+      "loss": 0.1116,
+      "step": 8393
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5301772497667172e-06,
+      "loss": 0.0841,
+      "step": 8394
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5279893231001108e-06,
+      "loss": 0.0669,
+      "step": 8395
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5258028778098277e-06,
+      "loss": 0.0666,
+      "step": 8396
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5236179141362904e-06,
+      "loss": 0.1225,
+      "step": 8397
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5214344323197555e-06,
+      "loss": 0.1064,
+      "step": 8398
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5192524326003122e-06,
+      "loss": 0.1587,
+      "step": 8399
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5170719152178968e-06,
+      "loss": 0.1148,
+      "step": 8400
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5148928804122813e-06,
+      "loss": 0.112,
+      "step": 8401
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5127153284230638e-06,
+      "loss": 0.0804,
+      "step": 8402
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5105392594896945e-06,
+      "loss": 0.0867,
+      "step": 8403
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5083646738514467e-06,
+      "loss": 0.0948,
+      "step": 8404
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5061915717474412e-06,
+      "loss": 0.0607,
+      "step": 8405
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5040199534166265e-06,
+      "loss": 0.056,
+      "step": 8406
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5018498190977986e-06,
+      "loss": 0.1134,
+      "step": 8407
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4996811690295797e-06,
+      "loss": 0.0592,
+      "step": 8408
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4975140034504391e-06,
+      "loss": 0.09,
+      "step": 8409
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4953483225986747e-06,
+      "loss": 0.0917,
+      "step": 8410
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.493184126712419e-06,
+      "loss": 0.1315,
+      "step": 8411
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4910214160296504e-06,
+      "loss": 0.1003,
+      "step": 8412
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4888601907881799e-06,
+      "loss": 0.0619,
+      "step": 8413
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4867004512256578e-06,
+      "loss": 0.0955,
+      "step": 8414
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4845421975795637e-06,
+      "loss": 0.0981,
+      "step": 8415
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.482385430087218e-06,
+      "loss": 0.0541,
+      "step": 8416
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4802301489857777e-06,
+      "loss": 0.081,
+      "step": 8417
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.47807635451224e-06,
+      "loss": 0.115,
+      "step": 8418
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.475924046903432e-06,
+      "loss": 0.0667,
+      "step": 8419
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4737732263960225e-06,
+      "loss": 0.1349,
+      "step": 8420
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4716238932265125e-06,
+      "loss": 0.0725,
+      "step": 8421
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4694760476312463e-06,
+      "loss": 0.1165,
+      "step": 8422
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4673296898463934e-06,
+      "loss": 0.0664,
+      "step": 8423
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4651848201079698e-06,
+      "loss": 0.0551,
+      "step": 8424
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.463041438651827e-06,
+      "loss": 0.051,
+      "step": 8425
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4608995457136481e-06,
+      "loss": 0.0955,
+      "step": 8426
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4587591415289514e-06,
+      "loss": 0.0779,
+      "step": 8427
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4566202263330968e-06,
+      "loss": 0.1061,
+      "step": 8428
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4544828003612798e-06,
+      "loss": 0.1004,
+      "step": 8429
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4523468638485338e-06,
+      "loss": 0.0429,
+      "step": 8430
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4502124170297194e-06,
+      "loss": 0.1096,
+      "step": 8431
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4480794601395436e-06,
+      "loss": 0.0876,
+      "step": 8432
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4459479934125424e-06,
+      "loss": 0.0358,
+      "step": 8433
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.443818017083095e-06,
+      "loss": 0.0851,
+      "step": 8434
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4416895313854056e-06,
+      "loss": 0.0666,
+      "step": 8435
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.439562536553527e-06,
+      "loss": 0.1022,
+      "step": 8436
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4374370328213455e-06,
+      "loss": 0.0834,
+      "step": 8437
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4353130204225757e-06,
+      "loss": 0.0697,
+      "step": 8438
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4331904995907708e-06,
+      "loss": 0.0755,
+      "step": 8439
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4310694705593259e-06,
+      "loss": 0.0959,
+      "step": 8440
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4289499335614676e-06,
+      "loss": 0.0946,
+      "step": 8441
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4268318888302629e-06,
+      "loss": 0.0779,
+      "step": 8442
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4247153365986071e-06,
+      "loss": 0.0763,
+      "step": 8443
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4226002770992324e-06,
+      "loss": 0.0969,
+      "step": 8444
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4204867105647146e-06,
+      "loss": 0.0987,
+      "step": 8445
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.418374637227459e-06,
+      "loss": 0.062,
+      "step": 8446
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4162640573197122e-06,
+      "loss": 0.0844,
+      "step": 8447
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.414154971073548e-06,
+      "loss": 0.1031,
+      "step": 8448
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4120473787208794e-06,
+      "loss": 0.0821,
+      "step": 8449
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4099412804934592e-06,
+      "loss": 0.0721,
+      "step": 8450
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4078366766228739e-06,
+      "loss": 0.1418,
+      "step": 8451
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4057335673405403e-06,
+      "loss": 0.1325,
+      "step": 8452
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4036319528777231e-06,
+      "loss": 0.0792,
+      "step": 8453
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4015318334655058e-06,
+      "loss": 0.0747,
+      "step": 8454
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3994332093348238e-06,
+      "loss": 0.1067,
+      "step": 8455
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3973360807164342e-06,
+      "loss": 0.0796,
+      "step": 8456
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3952404478409424e-06,
+      "loss": 0.0741,
+      "step": 8457
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3931463109387792e-06,
+      "loss": 0.0741,
+      "step": 8458
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3910536702402204e-06,
+      "loss": 0.0776,
+      "step": 8459
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3889625259753685e-06,
+      "loss": 0.0706,
+      "step": 8460
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3868728783741597e-06,
+      "loss": 0.1204,
+      "step": 8461
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3847847276663782e-06,
+      "loss": 0.0876,
+      "step": 8462
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3826980740816342e-06,
+      "loss": 0.1096,
+      "step": 8463
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3806129178493736e-06,
+      "loss": 0.0498,
+      "step": 8464
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3785292591988818e-06,
+      "loss": 0.0495,
+      "step": 8465
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.376447098359272e-06,
+      "loss": 0.1206,
+      "step": 8466
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3743664355595043e-06,
+      "loss": 0.1051,
+      "step": 8467
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3722872710283606e-06,
+      "loss": 0.0427,
+      "step": 8468
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3702096049944702e-06,
+      "loss": 0.0867,
+      "step": 8469
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3681334376862931e-06,
+      "loss": 0.0849,
+      "step": 8470
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.366058769332117e-06,
+      "loss": 0.1035,
+      "step": 8471
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3639856001600793e-06,
+      "loss": 0.046,
+      "step": 8472
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3619139303981376e-06,
+      "loss": 0.102,
+      "step": 8473
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3598437602740965e-06,
+      "loss": 0.1003,
+      "step": 8474
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3577750900155922e-06,
+      "loss": 0.0846,
+      "step": 8475
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3557079198500927e-06,
+      "loss": 0.0753,
+      "step": 8476
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3536422500048994e-06,
+      "loss": 0.1153,
+      "step": 8477
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3515780807071575e-06,
+      "loss": 0.0654,
+      "step": 8478
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3495154121838387e-06,
+      "loss": 0.0923,
+      "step": 8479
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3474542446617583e-06,
+      "loss": 0.0992,
+      "step": 8480
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3453945783675582e-06,
+      "loss": 0.0775,
+      "step": 8481
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3433364135277155e-06,
+      "loss": 0.0602,
+      "step": 8482
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3412797503685491e-06,
+      "loss": 0.0639,
+      "step": 8483
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3392245891162087e-06,
+      "loss": 0.0581,
+      "step": 8484
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3371709299966744e-06,
+      "loss": 0.1013,
+      "step": 8485
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3351187732357695e-06,
+      "loss": 0.082,
+      "step": 8486
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3330681190591516e-06,
+      "loss": 0.0965,
+      "step": 8487
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3310189676923051e-06,
+      "loss": 0.066,
+      "step": 8488
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.32897131936055e-06,
+      "loss": 0.0793,
+      "step": 8489
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3269251742890515e-06,
+      "loss": 0.0661,
+      "step": 8490
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3248805327028007e-06,
+      "loss": 0.1335,
+      "step": 8491
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3228373948266264e-06,
+      "loss": 0.0753,
+      "step": 8492
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3207957608851905e-06,
+      "loss": 0.1,
+      "step": 8493
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3187556311029885e-06,
+      "loss": 0.0559,
+      "step": 8494
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3167170057043526e-06,
+      "loss": 0.0365,
+      "step": 8495
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3146798849134517e-06,
+      "loss": 0.0633,
+      "step": 8496
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3126442689542834e-06,
+      "loss": 0.0944,
+      "step": 8497
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3106101580506886e-06,
+      "loss": 0.0906,
+      "step": 8498
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3085775524263284e-06,
+      "loss": 0.0963,
+      "step": 8499
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3065464523047155e-06,
+      "loss": 0.0232,
+      "step": 8500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3045168579091832e-06,
+      "loss": 0.1003,
+      "step": 8501
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3024887694629078e-06,
+      "loss": 0.1341,
+      "step": 8502
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3004621871888977e-06,
+      "loss": 0.0659,
+      "step": 8503
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2984371113099914e-06,
+      "loss": 0.0919,
+      "step": 8504
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2964135420488687e-06,
+      "loss": 0.07,
+      "step": 8505
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2943914796280387e-06,
+      "loss": 0.0957,
+      "step": 8506
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.292370924269845e-06,
+      "loss": 0.0474,
+      "step": 8507
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.29035187619647e-06,
+      "loss": 0.0944,
+      "step": 8508
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2883343356299292e-06,
+      "loss": 0.0957,
+      "step": 8509
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2863183027920673e-06,
+      "loss": 0.1032,
+      "step": 8510
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.284303777904563e-06,
+      "loss": 0.0982,
+      "step": 8511
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.282290761188938e-06,
+      "loss": 0.1062,
+      "step": 8512
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2802792528665413e-06,
+      "loss": 0.0875,
+      "step": 8513
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2782692531585566e-06,
+      "loss": 0.1267,
+      "step": 8514
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2762607622860034e-06,
+      "loss": 0.0633,
+      "step": 8515
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2742537804697334e-06,
+      "loss": 0.0995,
+      "step": 8516
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2722483079304353e-06,
+      "loss": 0.0932,
+      "step": 8517
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2702443448886274e-06,
+      "loss": 0.1121,
+      "step": 8518
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2682418915646654e-06,
+      "loss": 0.0619,
+      "step": 8519
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2662409481787412e-06,
+      "loss": 0.1238,
+      "step": 8520
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2642415149508757e-06,
+      "loss": 0.0777,
+      "step": 8521
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2622435921009228e-06,
+      "loss": 0.0462,
+      "step": 8522
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2602471798485737e-06,
+      "loss": 0.1111,
+      "step": 8523
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2582522784133576e-06,
+      "loss": 0.0894,
+      "step": 8524
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.256258888014631e-06,
+      "loss": 0.1257,
+      "step": 8525
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2542670088715863e-06,
+      "loss": 0.1029,
+      "step": 8526
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2522766412032455e-06,
+      "loss": 0.0926,
+      "step": 8527
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2502877852284733e-06,
+      "loss": 0.0731,
+      "step": 8528
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2483004411659632e-06,
+      "loss": 0.1266,
+      "step": 8529
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2463146092342404e-06,
+      "loss": 0.0485,
+      "step": 8530
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2443302896516684e-06,
+      "loss": 0.0828,
+      "step": 8531
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2423474826364394e-06,
+      "loss": 0.0821,
+      "step": 8532
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2403661884065826e-06,
+      "loss": 0.1219,
+      "step": 8533
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2383864071799634e-06,
+      "loss": 0.0781,
+      "step": 8534
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2364081391742744e-06,
+      "loss": 0.0456,
+      "step": 8535
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.234431384607045e-06,
+      "loss": 0.1122,
+      "step": 8536
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2324561436956412e-06,
+      "loss": 0.0751,
+      "step": 8537
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2304824166572593e-06,
+      "loss": 0.1465,
+      "step": 8538
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.228510203708924e-06,
+      "loss": 0.1024,
+      "step": 8539
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2265395050675037e-06,
+      "loss": 0.102,
+      "step": 8540
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.224570320949695e-06,
+      "loss": 0.0842,
+      "step": 8541
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2226026515720297e-06,
+      "loss": 0.0753,
+      "step": 8542
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.220636497150871e-06,
+      "loss": 0.1035,
+      "step": 8543
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2186718579024132e-06,
+      "loss": 0.0621,
+      "step": 8544
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2167087340426897e-06,
+      "loss": 0.1031,
+      "step": 8545
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2147471257875665e-06,
+      "loss": 0.0478,
+      "step": 8546
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2127870333527375e-06,
+      "loss": 0.0771,
+      "step": 8547
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2108284569537388e-06,
+      "loss": 0.0851,
+      "step": 8548
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.208871396805928e-06,
+      "loss": 0.0497,
+      "step": 8549
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2069158531245078e-06,
+      "loss": 0.0643,
+      "step": 8550
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2049618261245049e-06,
+      "loss": 0.0936,
+      "step": 8551
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2030093160207856e-06,
+      "loss": 0.0744,
+      "step": 8552
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2010583230280482e-06,
+      "loss": 0.055,
+      "step": 8553
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1991088473608226e-06,
+      "loss": 0.1048,
+      "step": 8554
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.197160889233469e-06,
+      "loss": 0.1021,
+      "step": 8555
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1952144488601845e-06,
+      "loss": 0.0815,
+      "step": 8556
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.193269526455003e-06,
+      "loss": 0.044,
+      "step": 8557
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1913261222317868e-06,
+      "loss": 0.0989,
+      "step": 8558
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1893842364042261e-06,
+      "loss": 0.0962,
+      "step": 8559
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1874438691858575e-06,
+      "loss": 0.0558,
+      "step": 8560
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1855050207900347e-06,
+      "loss": 0.0977,
+      "step": 8561
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1835676914299592e-06,
+      "loss": 0.0539,
+      "step": 8562
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1816318813186572e-06,
+      "loss": 0.0668,
+      "step": 8563
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1796975906689867e-06,
+      "loss": 0.1201,
+      "step": 8564
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1777648196936475e-06,
+      "loss": 0.1217,
+      "step": 8565
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.175833568605158e-06,
+      "loss": 0.0882,
+      "step": 8566
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1739038376158867e-06,
+      "loss": 0.0796,
+      "step": 8567
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1719756269380173e-06,
+      "loss": 0.0487,
+      "step": 8568
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1700489367835782e-06,
+      "loss": 0.0568,
+      "step": 8569
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1681237673644334e-06,
+      "loss": 0.095,
+      "step": 8570
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1662001188922672e-06,
+      "loss": 0.0738,
+      "step": 8571
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1642779915786018e-06,
+      "loss": 0.1127,
+      "step": 8572
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1623573856347946e-06,
+      "loss": 0.0587,
+      "step": 8573
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1604383012720387e-06,
+      "loss": 0.1566,
+      "step": 8574
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1585207387013552e-06,
+      "loss": 0.0805,
+      "step": 8575
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1566046981335953e-06,
+      "loss": 0.0999,
+      "step": 8576
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1546901797794456e-06,
+      "loss": 0.0557,
+      "step": 8577
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1527771838494262e-06,
+      "loss": 0.0892,
+      "step": 8578
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1508657105538933e-06,
+      "loss": 0.0482,
+      "step": 8579
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1489557601030258e-06,
+      "loss": 0.0902,
+      "step": 8580
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.147047332706846e-06,
+      "loss": 0.0675,
+      "step": 8581
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1451404285752003e-06,
+      "loss": 0.0801,
+      "step": 8582
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1432350479177734e-06,
+      "loss": 0.0668,
+      "step": 8583
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1413311909440771e-06,
+      "loss": 0.1174,
+      "step": 8584
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.139428857863461e-06,
+      "loss": 0.0763,
+      "step": 8585
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1375280488851041e-06,
+      "loss": 0.1105,
+      "step": 8586
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1356287642180213e-06,
+      "loss": 0.0642,
+      "step": 8587
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1337310040710536e-06,
+      "loss": 0.1331,
+      "step": 8588
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1318347686528774e-06,
+      "loss": 0.0704,
+      "step": 8589
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1299400581720026e-06,
+      "loss": 0.0809,
+      "step": 8590
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1280468728367743e-06,
+      "loss": 0.0998,
+      "step": 8591
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1261552128553608e-06,
+      "loss": 0.0642,
+      "step": 8592
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1242650784357745e-06,
+      "loss": 0.0718,
+      "step": 8593
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1223764697858468e-06,
+      "loss": 0.0893,
+      "step": 8594
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1204893871132505e-06,
+      "loss": 0.0791,
+      "step": 8595
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1186038306254926e-06,
+      "loss": 0.0805,
+      "step": 8596
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1167198005299039e-06,
+      "loss": 0.0908,
+      "step": 8597
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.114837297033654e-06,
+      "loss": 0.0698,
+      "step": 8598
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1129563203437371e-06,
+      "loss": 0.0631,
+      "step": 8599
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1110768706669927e-06,
+      "loss": 0.1159,
+      "step": 8600
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1091989482100757e-06,
+      "loss": 0.0432,
+      "step": 8601
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1073225531794856e-06,
+      "loss": 0.0858,
+      "step": 8602
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1054476857815525e-06,
+      "loss": 0.074,
+      "step": 8603
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.103574346222435e-06,
+      "loss": 0.0921,
+      "step": 8604
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.10170253470812e-06,
+      "loss": 0.13,
+      "step": 8605
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0998322514444325e-06,
+      "loss": 0.1165,
+      "step": 8606
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0979634966370317e-06,
+      "loss": 0.0823,
+      "step": 8607
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0960962704914047e-06,
+      "loss": 0.0655,
+      "step": 8608
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0942305732128705e-06,
+      "loss": 0.0871,
+      "step": 8609
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0923664050065752e-06,
+      "loss": 0.122,
+      "step": 8610
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0905037660775064e-06,
+      "loss": 0.0596,
+      "step": 8611
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0886426566304803e-06,
+      "loss": 0.1349,
+      "step": 8612
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0867830768701397e-06,
+      "loss": 0.1108,
+      "step": 8613
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0849250270009648e-06,
+      "loss": 0.0907,
+      "step": 8614
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.08306850722727e-06,
+      "loss": 0.0616,
+      "step": 8615
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.081213517753194e-06,
+      "loss": 0.0573,
+      "step": 8616
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0793600587827069e-06,
+      "loss": 0.1383,
+      "step": 8617
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0775081305196188e-06,
+      "loss": 0.1163,
+      "step": 8618
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0756577331675637e-06,
+      "loss": 0.0839,
+      "step": 8619
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0738088669300172e-06,
+      "loss": 0.1054,
+      "step": 8620
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.071961532010275e-06,
+      "loss": 0.101,
+      "step": 8621
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0701157286114676e-06,
+      "loss": 0.0753,
+      "step": 8622
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0682714569365599e-06,
+      "loss": 0.0802,
+      "step": 8623
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0664287171883475e-06,
+      "loss": 0.088,
+      "step": 8624
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.06458750956946e-06,
+      "loss": 0.0773,
+      "step": 8625
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0627478342823543e-06,
+      "loss": 0.0764,
+      "step": 8626
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0609096915293166e-06,
+      "loss": 0.107,
+      "step": 8627
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.05907308151247e-06,
+      "loss": 0.0779,
+      "step": 8628
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0572380044337732e-06,
+      "loss": 0.0609,
+      "step": 8629
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0554044604950014e-06,
+      "loss": 0.0912,
+      "step": 8630
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0535724498977783e-06,
+      "loss": 0.0912,
+      "step": 8631
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.051741972843543e-06,
+      "loss": 0.1265,
+      "step": 8632
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0499130295335823e-06,
+      "loss": 0.0506,
+      "step": 8633
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0480856201689976e-06,
+      "loss": 0.1303,
+      "step": 8634
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0462597449507362e-06,
+      "loss": 0.098,
+      "step": 8635
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0444354040795673e-06,
+      "loss": 0.0842,
+      "step": 8636
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0426125977560991e-06,
+      "loss": 0.0864,
+      "step": 8637
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0407913261807627e-06,
+      "loss": 0.0905,
+      "step": 8638
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0389715895538233e-06,
+      "loss": 0.112,
+      "step": 8639
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0371533880753803e-06,
+      "loss": 0.0559,
+      "step": 8640
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0353367219453641e-06,
+      "loss": 0.0772,
+      "step": 8641
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0335215913635298e-06,
+      "loss": 0.0742,
+      "step": 8642
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0317079965294729e-06,
+      "loss": 0.0925,
+      "step": 8643
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.029895937642612e-06,
+      "loss": 0.0811,
+      "step": 8644
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0280854149022046e-06,
+      "loss": 0.076,
+      "step": 8645
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0262764285073296e-06,
+      "loss": 0.0956,
+      "step": 8646
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.024468978656906e-06,
+      "loss": 0.1411,
+      "step": 8647
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0226630655496805e-06,
+      "loss": 0.1109,
+      "step": 8648
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0208586893842288e-06,
+      "loss": 0.1422,
+      "step": 8649
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0190558503589586e-06,
+      "loss": 0.106,
+      "step": 8650
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0172545486721086e-06,
+      "loss": 0.1116,
+      "step": 8651
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0154547845217515e-06,
+      "loss": 0.1156,
+      "step": 8652
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0136565581057911e-06,
+      "loss": 0.0234,
+      "step": 8653
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0118598696219572e-06,
+      "loss": 0.0838,
+      "step": 8654
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0100647192678086e-06,
+      "loss": 0.066,
+      "step": 8655
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0082711072407425e-06,
+      "loss": 0.0372,
+      "step": 8656
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0064790337379843e-06,
+      "loss": 0.0796,
+      "step": 8657
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.004688498956593e-06,
+      "loss": 0.1322,
+      "step": 8658
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0028995030934495e-06,
+      "loss": 0.0874,
+      "step": 8659
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0011120463452717e-06,
+      "loss": 0.095,
+      "step": 8660
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.99326128908607e-07,
+      "loss": 0.0762,
+      "step": 8661
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.975417509798401e-07,
+      "loss": 0.0805,
+      "step": 8662
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.957589127551725e-07,
+      "loss": 0.0717,
+      "step": 8663
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.939776144306473e-07,
+      "loss": 0.0614,
+      "step": 8664
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.921978562021394e-07,
+      "loss": 0.0706,
+      "step": 8665
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.904196382653457e-07,
+      "loss": 0.0811,
+      "step": 8666
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.886429608157981e-07,
+      "loss": 0.0536,
+      "step": 8667
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.868678240488588e-07,
+      "loss": 0.0933,
+      "step": 8668
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.850942281597247e-07,
+      "loss": 0.0607,
+      "step": 8669
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.833221733434183e-07,
+      "loss": 0.0522,
+      "step": 8670
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.815516597947938e-07,
+      "loss": 0.1013,
+      "step": 8671
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.79782687708532e-07,
+      "loss": 0.094,
+      "step": 8672
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.780152572791524e-07,
+      "loss": 0.1023,
+      "step": 8673
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.762493687010028e-07,
+      "loss": 0.0644,
+      "step": 8674
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.744850221682544e-07,
+      "loss": 0.0639,
+      "step": 8675
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.72722217874919e-07,
+      "loss": 0.0977,
+      "step": 8676
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.709609560148298e-07,
+      "loss": 0.0698,
+      "step": 8677
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.692012367816566e-07,
+      "loss": 0.1214,
+      "step": 8678
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.674430603688949e-07,
+      "loss": 0.1179,
+      "step": 8679
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.65686426969875e-07,
+      "loss": 0.0919,
+      "step": 8680
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.639313367777575e-07,
+      "loss": 0.0611,
+      "step": 8681
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.62177789985526e-07,
+      "loss": 0.0931,
+      "step": 8682
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.604257867860067e-07,
+      "loss": 0.0833,
+      "step": 8683
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.586753273718417e-07,
+      "loss": 0.0934,
+      "step": 8684
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.569264119355143e-07,
+      "loss": 0.051,
+      "step": 8685
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.55179040669335e-07,
+      "loss": 0.1146,
+      "step": 8686
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.534332137654472e-07,
+      "loss": 0.0452,
+      "step": 8687
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.516889314158139e-07,
+      "loss": 0.1169,
+      "step": 8688
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.499461938122384e-07,
+      "loss": 0.0527,
+      "step": 8689
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.482050011463523e-07,
+      "loss": 0.0473,
+      "step": 8690
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.464653536096191e-07,
+      "loss": 0.0751,
+      "step": 8691
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.447272513933242e-07,
+      "loss": 0.0877,
+      "step": 8692
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.42990694688593e-07,
+      "loss": 0.1101,
+      "step": 8693
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.412556836863712e-07,
+      "loss": 0.069,
+      "step": 8694
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.395222185774476e-07,
+      "loss": 0.0996,
+      "step": 8695
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.377902995524252e-07,
+      "loss": 0.0486,
+      "step": 8696
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.360599268017494e-07,
+      "loss": 0.0572,
+      "step": 8697
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.343311005156918e-07,
+      "loss": 0.0588,
+      "step": 8698
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.326038208843535e-07,
+      "loss": 0.1008,
+      "step": 8699
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.308780880976625e-07,
+      "loss": 0.0586,
+      "step": 8700
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.291539023453787e-07,
+      "loss": 0.0834,
+      "step": 8701
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.274312638170973e-07,
+      "loss": 0.097,
+      "step": 8702
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.257101727022382e-07,
+      "loss": 0.0725,
+      "step": 8703
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.239906291900502e-07,
+      "loss": 0.0877,
+      "step": 8704
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.222726334696119e-07,
+      "loss": 0.1038,
+      "step": 8705
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.205561857298356e-07,
+      "loss": 0.0927,
+      "step": 8706
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.188412861594619e-07,
+      "loss": 0.0696,
+      "step": 8707
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.171279349470585e-07,
+      "loss": 0.0995,
+      "step": 8708
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.154161322810279e-07,
+      "loss": 0.0993,
+      "step": 8709
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.137058783495927e-07,
+      "loss": 0.0876,
+      "step": 8710
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.119971733408194e-07,
+      "loss": 0.1401,
+      "step": 8711
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.102900174425893e-07,
+      "loss": 0.1061,
+      "step": 8712
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.085844108426256e-07,
+      "loss": 0.0855,
+      "step": 8713
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.068803537284732e-07,
+      "loss": 0.1167,
+      "step": 8714
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.051778462875137e-07,
+      "loss": 0.1269,
+      "step": 8715
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.034768887069511e-07,
+      "loss": 0.095,
+      "step": 8716
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.017774811738188e-07,
+      "loss": 0.1071,
+      "step": 8717
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.000796238749858e-07,
+      "loss": 0.0435,
+      "step": 8718
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.983833169971495e-07,
+      "loss": 0.0818,
+      "step": 8719
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.966885607268338e-07,
+      "loss": 0.0703,
+      "step": 8720
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.949953552503948e-07,
+      "loss": 0.1305,
+      "step": 8721
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.933037007540117e-07,
+      "loss": 0.0812,
+      "step": 8722
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.916135974237011e-07,
+      "loss": 0.0474,
+      "step": 8723
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.89925045445309e-07,
+      "loss": 0.0996,
+      "step": 8724
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.88238045004502e-07,
+      "loss": 0.0895,
+      "step": 8725
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.865525962867882e-07,
+      "loss": 0.0659,
+      "step": 8726
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.84868699477493e-07,
+      "loss": 0.0307,
+      "step": 8727
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.831863547617813e-07,
+      "loss": 0.0705,
+      "step": 8728
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.815055623246387e-07,
+      "loss": 0.0849,
+      "step": 8729
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.798263223508873e-07,
+      "loss": 0.0679,
+      "step": 8730
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.781486350251761e-07,
+      "loss": 0.0826,
+      "step": 8731
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.76472500531984e-07,
+      "loss": 0.1101,
+      "step": 8732
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.747979190556121e-07,
+      "loss": 0.0845,
+      "step": 8733
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.731248907801997e-07,
+      "loss": 0.0979,
+      "step": 8734
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.714534158897148e-07,
+      "loss": 0.0659,
+      "step": 8735
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.697834945679517e-07,
+      "loss": 0.0615,
+      "step": 8736
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.681151269985321e-07,
+      "loss": 0.0609,
+      "step": 8737
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.664483133649076e-07,
+      "loss": 0.1071,
+      "step": 8738
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.647830538503632e-07,
+      "loss": 0.1299,
+      "step": 8739
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.631193486380107e-07,
+      "loss": 0.1261,
+      "step": 8740
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.614571979107855e-07,
+      "loss": 0.12,
+      "step": 8741
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.597966018514614e-07,
+      "loss": 0.0506,
+      "step": 8742
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.581375606426373e-07,
+      "loss": 0.0757,
+      "step": 8743
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.564800744667373e-07,
+      "loss": 0.0861,
+      "step": 8744
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.548241435060222e-07,
+      "loss": 0.0667,
+      "step": 8745
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.531697679425716e-07,
+      "loss": 0.0862,
+      "step": 8746
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.515169479583046e-07,
+      "loss": 0.0895,
+      "step": 8747
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.498656837349644e-07,
+      "loss": 0.0726,
+      "step": 8748
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.482159754541207e-07,
+      "loss": 0.0677,
+      "step": 8749
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.465678232971752e-07,
+      "loss": 0.1048,
+      "step": 8750
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.449212274453577e-07,
+      "loss": 0.0866,
+      "step": 8751
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.432761880797302e-07,
+      "loss": 0.1386,
+      "step": 8752
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.416327053811779e-07,
+      "loss": 0.058,
+      "step": 8753
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.399907795304196e-07,
+      "loss": 0.0504,
+      "step": 8754
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.383504107079976e-07,
+      "loss": 0.0788,
+      "step": 8755
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.367115990942859e-07,
+      "loss": 0.0519,
+      "step": 8756
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.35074344869492e-07,
+      "loss": 0.0849,
+      "step": 8757
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.334386482136436e-07,
+      "loss": 0.0722,
+      "step": 8758
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.318045093066034e-07,
+      "loss": 0.0927,
+      "step": 8759
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.301719283280596e-07,
+      "loss": 0.0757,
+      "step": 8760
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.2854090545753e-07,
+      "loss": 0.1077,
+      "step": 8761
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.269114408743611e-07,
+      "loss": 0.0493,
+      "step": 8762
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.252835347577264e-07,
+      "loss": 0.0868,
+      "step": 8763
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.236571872866327e-07,
+      "loss": 0.1353,
+      "step": 8764
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.220323986399153e-07,
+      "loss": 0.1208,
+      "step": 8765
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.204091689962279e-07,
+      "loss": 0.0463,
+      "step": 8766
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.187874985340626e-07,
+      "loss": 0.0868,
+      "step": 8767
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.171673874317387e-07,
+      "loss": 0.051,
+      "step": 8768
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.155488358674051e-07,
+      "loss": 0.0864,
+      "step": 8769
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.139318440190313e-07,
+      "loss": 0.0469,
+      "step": 8770
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.123164120644283e-07,
+      "loss": 0.0477,
+      "step": 8771
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.107025401812207e-07,
+      "loss": 0.1089,
+      "step": 8772
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.090902285468732e-07,
+      "loss": 0.0764,
+      "step": 8773
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.074794773386773e-07,
+      "loss": 0.0356,
+      "step": 8774
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.058702867337447e-07,
+      "loss": 0.0475,
+      "step": 8775
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.042626569090272e-07,
+      "loss": 0.0666,
+      "step": 8776
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.026565880412934e-07,
+      "loss": 0.0732,
+      "step": 8777
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.010520803071503e-07,
+      "loss": 0.1085,
+      "step": 8778
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.994491338830251e-07,
+      "loss": 0.0429,
+      "step": 8779
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.978477489451785e-07,
+      "loss": 0.1183,
+      "step": 8780
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.962479256697014e-07,
+      "loss": 0.0597,
+      "step": 8781
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.946496642325079e-07,
+      "loss": 0.0599,
+      "step": 8782
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.930529648093377e-07,
+      "loss": 0.105,
+      "step": 8783
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.914578275757684e-07,
+      "loss": 0.1093,
+      "step": 8784
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.898642527071965e-07,
+      "loss": 0.0853,
+      "step": 8785
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.882722403788567e-07,
+      "loss": 0.047,
+      "step": 8786
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.866817907658025e-07,
+      "loss": 0.0608,
+      "step": 8787
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.850929040429156e-07,
+      "loss": 0.0745,
+      "step": 8788
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.835055803849144e-07,
+      "loss": 0.066,
+      "step": 8789
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.819198199663379e-07,
+      "loss": 0.0446,
+      "step": 8790
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.803356229615566e-07,
+      "loss": 0.0897,
+      "step": 8791
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.78752989544766e-07,
+      "loss": 0.0575,
+      "step": 8792
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.771719198899968e-07,
+      "loss": 0.0983,
+      "step": 8793
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.755924141710985e-07,
+      "loss": 0.1015,
+      "step": 8794
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.740144725617537e-07,
+      "loss": 0.0783,
+      "step": 8795
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.724380952354704e-07,
+      "loss": 0.0704,
+      "step": 8796
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.708632823655881e-07,
+      "loss": 0.1263,
+      "step": 8797
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.692900341252751e-07,
+      "loss": 0.1144,
+      "step": 8798
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.677183506875212e-07,
+      "loss": 0.073,
+      "step": 8799
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.661482322251484e-07,
+      "loss": 0.1084,
+      "step": 8800
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.645796789108084e-07,
+      "loss": 0.1494,
+      "step": 8801
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.630126909169782e-07,
+      "loss": 0.0744,
+      "step": 8802
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.614472684159602e-07,
+      "loss": 0.0919,
+      "step": 8803
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.598834115798913e-07,
+      "loss": 0.0756,
+      "step": 8804
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.583211205807295e-07,
+      "loss": 0.1143,
+      "step": 8805
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.567603955902652e-07,
+      "loss": 0.0472,
+      "step": 8806
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.552012367801181e-07,
+      "loss": 0.1069,
+      "step": 8807
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.53643644321726e-07,
+      "loss": 0.1008,
+      "step": 8808
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.520876183863651e-07,
+      "loss": 0.0927,
+      "step": 8809
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.505331591451353e-07,
+      "loss": 0.0671,
+      "step": 8810
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.489802667689632e-07,
+      "loss": 0.1118,
+      "step": 8811
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.47428941428604e-07,
+      "loss": 0.1203,
+      "step": 8812
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.458791832946409e-07,
+      "loss": 0.0909,
+      "step": 8813
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.443309925374848e-07,
+      "loss": 0.0731,
+      "step": 8814
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.427843693273789e-07,
+      "loss": 0.0503,
+      "step": 8815
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.412393138343809e-07,
+      "loss": 0.1006,
+      "step": 8816
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.396958262283881e-07,
+      "loss": 0.0877,
+      "step": 8817
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.381539066791198e-07,
+      "loss": 0.0949,
+      "step": 8818
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.366135553561304e-07,
+      "loss": 0.1022,
+      "step": 8819
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.35074772428791e-07,
+      "loss": 0.054,
+      "step": 8820
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.335375580663079e-07,
+      "loss": 0.0928,
+      "step": 8821
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.320019124377108e-07,
+      "loss": 0.1242,
+      "step": 8822
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.304678357118616e-07,
+      "loss": 0.0929,
+      "step": 8823
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.289353280574418e-07,
+      "loss": 0.0977,
+      "step": 8824
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.274043896429667e-07,
+      "loss": 0.0588,
+      "step": 8825
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.258750206367831e-07,
+      "loss": 0.0448,
+      "step": 8826
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.243472212070551e-07,
+      "loss": 0.0637,
+      "step": 8827
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.228209915217765e-07,
+      "loss": 0.0671,
+      "step": 8828
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.212963317487747e-07,
+      "loss": 0.052,
+      "step": 8829
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.197732420556991e-07,
+      "loss": 0.0701,
+      "step": 8830
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.182517226100304e-07,
+      "loss": 0.1197,
+      "step": 8831
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.167317735790735e-07,
+      "loss": 0.0761,
+      "step": 8832
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.152133951299561e-07,
+      "loss": 0.0737,
+      "step": 8833
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.13696587429643e-07,
+      "loss": 0.0607,
+      "step": 8834
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.121813506449227e-07,
+      "loss": 0.1005,
+      "step": 8835
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.106676849424082e-07,
+      "loss": 0.097,
+      "step": 8836
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.091555904885433e-07,
+      "loss": 0.0827,
+      "step": 8837
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.076450674495932e-07,
+      "loss": 0.125,
+      "step": 8838
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.061361159916552e-07,
+      "loss": 0.1034,
+      "step": 8839
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.046287362806581e-07,
+      "loss": 0.1462,
+      "step": 8840
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.031229284823476e-07,
+      "loss": 0.0886,
+      "step": 8841
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.016186927623014e-07,
+      "loss": 0.0907,
+      "step": 8842
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.001160292859305e-07,
+      "loss": 0.0705,
+      "step": 8843
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.986149382184609e-07,
+      "loss": 0.1158,
+      "step": 8844
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.971154197249541e-07,
+      "loss": 0.0725,
+      "step": 8845
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.956174739702947e-07,
+      "loss": 0.0775,
+      "step": 8846
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.941211011191978e-07,
+      "loss": 0.077,
+      "step": 8847
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.926263013362066e-07,
+      "loss": 0.0585,
+      "step": 8848
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.911330747856864e-07,
+      "loss": 0.1009,
+      "step": 8849
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.896414216318275e-07,
+      "loss": 0.0434,
+      "step": 8850
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.88151342038657e-07,
+      "loss": 0.0792,
+      "step": 8851
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.866628361700223e-07,
+      "loss": 0.106,
+      "step": 8852
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.851759041895972e-07,
+      "loss": 0.0727,
+      "step": 8853
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.836905462608861e-07,
+      "loss": 0.0448,
+      "step": 8854
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.82206762547215e-07,
+      "loss": 0.1065,
+      "step": 8855
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.807245532117451e-07,
+      "loss": 0.101,
+      "step": 8856
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.792439184174542e-07,
+      "loss": 0.084,
+      "step": 8857
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.777648583271557e-07,
+      "loss": 0.09,
+      "step": 8858
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.762873731034874e-07,
+      "loss": 0.1031,
+      "step": 8859
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.748114629089097e-07,
+      "loss": 0.0878,
+      "step": 8860
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.733371279057127e-07,
+      "loss": 0.099,
+      "step": 8861
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.71864368256015e-07,
+      "loss": 0.0813,
+      "step": 8862
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.70393184121762e-07,
+      "loss": 0.0856,
+      "step": 8863
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.689235756647244e-07,
+      "loss": 0.0759,
+      "step": 8864
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.674555430464979e-07,
+      "loss": 0.0961,
+      "step": 8865
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.659890864285067e-07,
+      "loss": 0.0864,
+      "step": 8866
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.645242059720002e-07,
+      "loss": 0.0338,
+      "step": 8867
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.630609018380596e-07,
+      "loss": 0.0754,
+      "step": 8868
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.615991741875909e-07,
+      "loss": 0.0707,
+      "step": 8869
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.601390231813193e-07,
+      "loss": 0.0359,
+      "step": 8870
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.58680448979806e-07,
+      "loss": 0.0772,
+      "step": 8871
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.572234517434327e-07,
+      "loss": 0.1147,
+      "step": 8872
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.557680316324132e-07,
+      "loss": 0.0856,
+      "step": 8873
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.54314188806781e-07,
+      "loss": 0.1297,
+      "step": 8874
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.528619234264032e-07,
+      "loss": 0.072,
+      "step": 8875
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.514112356509705e-07,
+      "loss": 0.0962,
+      "step": 8876
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.499621256399985e-07,
+      "loss": 0.0524,
+      "step": 8877
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.485145935528297e-07,
+      "loss": 0.0527,
+      "step": 8878
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.470686395486353e-07,
+      "loss": 0.0474,
+      "step": 8879
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.456242637864113e-07,
+      "loss": 0.0417,
+      "step": 8880
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.441814664249823e-07,
+      "loss": 0.051,
+      "step": 8881
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.427402476229982e-07,
+      "loss": 0.0557,
+      "step": 8882
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.413006075389288e-07,
+      "loss": 0.0896,
+      "step": 8883
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.398625463310825e-07,
+      "loss": 0.0686,
+      "step": 8884
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.384260641575878e-07,
+      "loss": 0.1046,
+      "step": 8885
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.369911611763935e-07,
+      "loss": 0.1121,
+      "step": 8886
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.355578375452897e-07,
+      "loss": 0.0762,
+      "step": 8887
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.341260934218757e-07,
+      "loss": 0.1029,
+      "step": 8888
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.326959289635919e-07,
+      "loss": 0.1035,
+      "step": 8889
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.312673443276928e-07,
+      "loss": 0.1355,
+      "step": 8890
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.298403396712693e-07,
+      "loss": 0.0923,
+      "step": 8891
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.284149151512325e-07,
+      "loss": 0.0552,
+      "step": 8892
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.269910709243254e-07,
+      "loss": 0.08,
+      "step": 8893
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.25568807147106e-07,
+      "loss": 0.0765,
+      "step": 8894
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.241481239759694e-07,
+      "loss": 0.0894,
+      "step": 8895
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.227290215671339e-07,
+      "loss": 0.0867,
+      "step": 8896
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.213115000766428e-07,
+      "loss": 0.0736,
+      "step": 8897
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.198955596603667e-07,
+      "loss": 0.1396,
+      "step": 8898
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.184812004740026e-07,
+      "loss": 0.0685,
+      "step": 8899
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.170684226730694e-07,
+      "loss": 0.0598,
+      "step": 8900
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.156572264129162e-07,
+      "loss": 0.1115,
+      "step": 8901
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.142476118487222e-07,
+      "loss": 0.124,
+      "step": 8902
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.128395791354835e-07,
+      "loss": 0.0757,
+      "step": 8903
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.114331284280278e-07,
+      "loss": 0.0711,
+      "step": 8904
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.100282598810081e-07,
+      "loss": 0.0948,
+      "step": 8905
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.086249736489025e-07,
+      "loss": 0.0483,
+      "step": 8906
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.072232698860158e-07,
+      "loss": 0.067,
+      "step": 8907
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.0582314874648e-07,
+      "loss": 0.0645,
+      "step": 8908
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.044246103842515e-07,
+      "loss": 0.1346,
+      "step": 8909
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.030276549531127e-07,
+      "loss": 0.079,
+      "step": 8910
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.016322826066689e-07,
+      "loss": 0.1471,
+      "step": 8911
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.00238493498359e-07,
+      "loss": 0.1082,
+      "step": 8912
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.988462877814421e-07,
+      "loss": 0.099,
+      "step": 8913
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.974556656090058e-07,
+      "loss": 0.0766,
+      "step": 8914
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.96066627133961e-07,
+      "loss": 0.0824,
+      "step": 8915
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.946791725090439e-07,
+      "loss": 0.0939,
+      "step": 8916
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.932933018868209e-07,
+      "loss": 0.1221,
+      "step": 8917
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.919090154196815e-07,
+      "loss": 0.0904,
+      "step": 8918
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.905263132598393e-07,
+      "loss": 0.0789,
+      "step": 8919
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.891451955593374e-07,
+      "loss": 0.0908,
+      "step": 8920
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.877656624700445e-07,
+      "loss": 0.0796,
+      "step": 8921
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.863877141436525e-07,
+      "loss": 0.1095,
+      "step": 8922
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.850113507316751e-07,
+      "loss": 0.0779,
+      "step": 8923
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.836365723854614e-07,
+      "loss": 0.0564,
+      "step": 8924
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.822633792561804e-07,
+      "loss": 0.1053,
+      "step": 8925
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.808917714948298e-07,
+      "loss": 0.0788,
+      "step": 8926
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.795217492522304e-07,
+      "loss": 0.0574,
+      "step": 8927
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.781533126790251e-07,
+      "loss": 0.1045,
+      "step": 8928
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.767864619256885e-07,
+      "loss": 0.0675,
+      "step": 8929
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.754211971425205e-07,
+      "loss": 0.1001,
+      "step": 8930
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.74057518479646e-07,
+      "loss": 0.0853,
+      "step": 8931
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.726954260870132e-07,
+      "loss": 0.17,
+      "step": 8932
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.713349201143958e-07,
+      "loss": 0.0751,
+      "step": 8933
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.699760007113941e-07,
+      "loss": 0.0491,
+      "step": 8934
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.686186680274386e-07,
+      "loss": 0.038,
+      "step": 8935
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.672629222117748e-07,
+      "loss": 0.0883,
+      "step": 8936
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.659087634134851e-07,
+      "loss": 0.0542,
+      "step": 8937
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.645561917814707e-07,
+      "loss": 0.0836,
+      "step": 8938
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.632052074644589e-07,
+      "loss": 0.1038,
+      "step": 8939
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.618558106110027e-07,
+      "loss": 0.076,
+      "step": 8940
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.605080013694819e-07,
+      "loss": 0.0494,
+      "step": 8941
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.591617798881026e-07,
+      "loss": 0.0716,
+      "step": 8942
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.578171463148951e-07,
+      "loss": 0.0507,
+      "step": 8943
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.564741007977092e-07,
+      "loss": 0.1367,
+      "step": 8944
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.551326434842318e-07,
+      "loss": 0.1348,
+      "step": 8945
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.53792774521965e-07,
+      "loss": 0.0703,
+      "step": 8946
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.524544940582443e-07,
+      "loss": 0.0772,
+      "step": 8947
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.511178022402219e-07,
+      "loss": 0.1168,
+      "step": 8948
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.497826992148835e-07,
+      "loss": 0.077,
+      "step": 8949
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.484491851290335e-07,
+      "loss": 0.078,
+      "step": 8950
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.471172601293062e-07,
+      "loss": 0.0913,
+      "step": 8951
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.457869243621593e-07,
+      "loss": 0.113,
+      "step": 8952
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.444581779738744e-07,
+      "loss": 0.089,
+      "step": 8953
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.431310211105628e-07,
+      "loss": 0.0573,
+      "step": 8954
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.418054539181561e-07,
+      "loss": 0.0786,
+      "step": 8955
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.404814765424143e-07,
+      "loss": 0.0733,
+      "step": 8956
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.391590891289195e-07,
+      "loss": 0.0891,
+      "step": 8957
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.378382918230801e-07,
+      "loss": 0.0994,
+      "step": 8958
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.365190847701351e-07,
+      "loss": 0.0751,
+      "step": 8959
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.352014681151419e-07,
+      "loss": 0.0893,
+      "step": 8960
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.338854420029809e-07,
+      "loss": 0.1522,
+      "step": 8961
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.325710065783651e-07,
+      "loss": 0.0513,
+      "step": 8962
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.312581619858304e-07,
+      "loss": 0.0568,
+      "step": 8963
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.299469083697361e-07,
+      "loss": 0.0789,
+      "step": 8964
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.286372458742655e-07,
+      "loss": 0.0519,
+      "step": 8965
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.273291746434283e-07,
+      "loss": 0.1028,
+      "step": 8966
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.260226948210611e-07,
+      "loss": 0.0821,
+      "step": 8967
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.24717806550824e-07,
+      "loss": 0.1097,
+      "step": 8968
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.234145099762005e-07,
+      "loss": 0.0859,
+      "step": 8969
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.221128052404995e-07,
+      "loss": 0.0712,
+      "step": 8970
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.208126924868611e-07,
+      "loss": 0.1048,
+      "step": 8971
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.195141718582397e-07,
+      "loss": 0.0983,
+      "step": 8972
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.182172434974208e-07,
+      "loss": 0.0822,
+      "step": 8973
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.169219075470155e-07,
+      "loss": 0.0559,
+      "step": 8974
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.156281641494581e-07,
+      "loss": 0.0944,
+      "step": 8975
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.143360134470082e-07,
+      "loss": 0.0767,
+      "step": 8976
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.130454555817521e-07,
+      "loss": 0.0885,
+      "step": 8977
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.117564906955929e-07,
+      "loss": 0.0885,
+      "step": 8978
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.104691189302707e-07,
+      "loss": 0.0849,
+      "step": 8979
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.091833404273422e-07,
+      "loss": 0.0781,
+      "step": 8980
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.078991553281892e-07,
+      "loss": 0.067,
+      "step": 8981
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.066165637740239e-07,
+      "loss": 0.0788,
+      "step": 8982
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.053355659058751e-07,
+      "loss": 0.0329,
+      "step": 8983
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.040561618646033e-07,
+      "loss": 0.1069,
+      "step": 8984
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.027783517908946e-07,
+      "loss": 0.1306,
+      "step": 8985
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.015021358252497e-07,
+      "loss": 0.0838,
+      "step": 8986
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.002275141080048e-07,
+      "loss": 0.1002,
+      "step": 8987
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.989544867793144e-07,
+      "loss": 0.092,
+      "step": 8988
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.976830539791633e-07,
+      "loss": 0.0577,
+      "step": 8989
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.964132158473544e-07,
+      "loss": 0.1001,
+      "step": 8990
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.951449725235196e-07,
+      "loss": 0.0706,
+      "step": 8991
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.938783241471173e-07,
+      "loss": 0.0812,
+      "step": 8992
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.926132708574244e-07,
+      "loss": 0.0634,
+      "step": 8993
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.913498127935445e-07,
+      "loss": 0.0963,
+      "step": 8994
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.900879500944083e-07,
+      "loss": 0.0784,
+      "step": 8995
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.888276828987697e-07,
+      "loss": 0.0889,
+      "step": 8996
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.875690113452097e-07,
+      "loss": 0.0603,
+      "step": 8997
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.863119355721273e-07,
+      "loss": 0.0793,
+      "step": 8998
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.850564557177539e-07,
+      "loss": 0.1035,
+      "step": 8999
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.838025719201372e-07,
+      "loss": 0.1371,
+      "step": 9000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.82550284317157e-07,
+      "loss": 0.0904,
+      "step": 9001
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.812995930465114e-07,
+      "loss": 0.1031,
+      "step": 9002
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.800504982457271e-07,
+      "loss": 0.0822,
+      "step": 9003
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.788030000521576e-07,
+      "loss": 0.0899,
+      "step": 9004
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.775570986029731e-07,
+      "loss": 0.0856,
+      "step": 9005
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.763127940351708e-07,
+      "loss": 0.075,
+      "step": 9006
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.750700864855778e-07,
+      "loss": 0.0304,
+      "step": 9007
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.738289760908399e-07,
+      "loss": 0.0952,
+      "step": 9008
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.725894629874311e-07,
+      "loss": 0.1093,
+      "step": 9009
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.7135154731164586e-07,
+      "loss": 0.0573,
+      "step": 9010
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.701152291996036e-07,
+      "loss": 0.1111,
+      "step": 9011
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.688805087872505e-07,
+      "loss": 0.0578,
+      "step": 9012
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6764738621035807e-07,
+      "loss": 0.0868,
+      "step": 9013
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.664158616045161e-07,
+      "loss": 0.0873,
+      "step": 9014
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.651859351051463e-07,
+      "loss": 0.0728,
+      "step": 9015
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.639576068474871e-07,
+      "loss": 0.0609,
+      "step": 9016
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6273087696660553e-07,
+      "loss": 0.1109,
+      "step": 9017
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.61505745597397e-07,
+      "loss": 0.1252,
+      "step": 9018
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.602822128745704e-07,
+      "loss": 0.1032,
+      "step": 9019
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.590602789326681e-07,
+      "loss": 0.0905,
+      "step": 9020
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.5783994390605265e-07,
+      "loss": 0.0304,
+      "step": 9021
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.566212079289117e-07,
+      "loss": 0.0808,
+      "step": 9022
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.554040711352564e-07,
+      "loss": 0.0891,
+      "step": 9023
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.54188533658923e-07,
+      "loss": 0.0941,
+      "step": 9024
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.529745956335696e-07,
+      "loss": 0.0505,
+      "step": 9025
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.5176225719268607e-07,
+      "loss": 0.0894,
+      "step": 9026
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.505515184695741e-07,
+      "loss": 0.0571,
+      "step": 9027
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.493423795973689e-07,
+      "loss": 0.0722,
+      "step": 9028
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.481348407090258e-07,
+      "loss": 0.0522,
+      "step": 9029
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.4692890193732685e-07,
+      "loss": 0.059,
+      "step": 9030
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.4572456341487433e-07,
+      "loss": 0.0911,
+      "step": 9031
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.4452182527410067e-07,
+      "loss": 0.0706,
+      "step": 9032
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.4332068764725343e-07,
+      "loss": 0.0578,
+      "step": 9033
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.421211506664119e-07,
+      "loss": 0.0521,
+      "step": 9034
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.409232144634756e-07,
+      "loss": 0.0984,
+      "step": 9035
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3972687917017087e-07,
+      "loss": 0.0694,
+      "step": 9036
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.385321449180457e-07,
+      "loss": 0.0537,
+      "step": 9037
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.373390118384718e-07,
+      "loss": 0.0567,
+      "step": 9038
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3614748006264414e-07,
+      "loss": 0.096,
+      "step": 9039
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3495754972158464e-07,
+      "loss": 0.1256,
+      "step": 9040
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.337692209461369e-07,
+      "loss": 0.1415,
+      "step": 9041
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.325824938669715e-07,
+      "loss": 0.031,
+      "step": 9042
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3139736861457724e-07,
+      "loss": 0.0436,
+      "step": 9043
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3021384531927166e-07,
+      "loss": 0.0856,
+      "step": 9044
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.290319241111923e-07,
+      "loss": 0.0754,
+      "step": 9045
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.278516051203052e-07,
+      "loss": 0.1034,
+      "step": 9046
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.2667288847639816e-07,
+      "loss": 0.1067,
+      "step": 9047
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.2549577430907927e-07,
+      "loss": 0.0956,
+      "step": 9048
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.2432026274778824e-07,
+      "loss": 0.0931,
+      "step": 9049
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.2314635392177834e-07,
+      "loss": 0.062,
+      "step": 9050
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.2197404796013474e-07,
+      "loss": 0.0795,
+      "step": 9051
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.208033449917642e-07,
+      "loss": 0.0978,
+      "step": 9052
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.1963424514539384e-07,
+      "loss": 0.0635,
+      "step": 9053
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.184667485495824e-07,
+      "loss": 0.0649,
+      "step": 9054
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.17300855332704e-07,
+      "loss": 0.0667,
+      "step": 9055
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.1613656562295933e-07,
+      "loss": 0.0577,
+      "step": 9056
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.1497387954837427e-07,
+      "loss": 0.0769,
+      "step": 9057
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.13812797236795e-07,
+      "loss": 0.0747,
+      "step": 9058
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.126533188158993e-07,
+      "loss": 0.1185,
+      "step": 9059
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.114954444131769e-07,
+      "loss": 0.0701,
+      "step": 9060
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.103391741559492e-07,
+      "loss": 0.0889,
+      "step": 9061
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0918450817135965e-07,
+      "loss": 0.0566,
+      "step": 9062
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.080314465863766e-07,
+      "loss": 0.0873,
+      "step": 9063
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.068799895277853e-07,
+      "loss": 0.1119,
+      "step": 9064
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.057301371222044e-07,
+      "loss": 0.0807,
+      "step": 9065
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0458188949606787e-07,
+      "loss": 0.0684,
+      "step": 9066
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.034352467756397e-07,
+      "loss": 0.1202,
+      "step": 9067
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0229020908700066e-07,
+      "loss": 0.1157,
+      "step": 9068
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0114677655605836e-07,
+      "loss": 0.0855,
+      "step": 9069
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.000049493085489e-07,
+      "loss": 0.1294,
+      "step": 9070
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.9886472747002357e-07,
+      "loss": 0.0828,
+      "step": 9071
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.977261111658603e-07,
+      "loss": 0.0565,
+      "step": 9072
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.965891005212624e-07,
+      "loss": 0.0896,
+      "step": 9073
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.954536956612531e-07,
+      "loss": 0.0844,
+      "step": 9074
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.943198967106826e-07,
+      "loss": 0.1368,
+      "step": 9075
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.9318770379422443e-07,
+      "loss": 0.1232,
+      "step": 9076
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.920571170363724e-07,
+      "loss": 0.0755,
+      "step": 9077
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.9092813656144365e-07,
+      "loss": 0.1435,
+      "step": 9078
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.898007624935823e-07,
+      "loss": 0.0806,
+      "step": 9079
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8867499495675416e-07,
+      "loss": 0.0601,
+      "step": 9080
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.875508340747469e-07,
+      "loss": 0.0451,
+      "step": 9081
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8642827997117325e-07,
+      "loss": 0.0599,
+      "step": 9082
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.853073327694695e-07,
+      "loss": 0.0888,
+      "step": 9083
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.841879925928937e-07,
+      "loss": 0.0592,
+      "step": 9084
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8307025956452744e-07,
+      "loss": 0.1095,
+      "step": 9085
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.819541338072757e-07,
+      "loss": 0.0772,
+      "step": 9086
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8083961544387205e-07,
+      "loss": 0.1145,
+      "step": 9087
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7972670459686334e-07,
+      "loss": 0.0768,
+      "step": 9088
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7861540138862505e-07,
+      "loss": 0.087,
+      "step": 9089
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7750570594135614e-07,
+      "loss": 0.0786,
+      "step": 9090
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.76397618377079e-07,
+      "loss": 0.0987,
+      "step": 9091
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7529113881763944e-07,
+      "loss": 0.0787,
+      "step": 9092
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.741862673847035e-07,
+      "loss": 0.0614,
+      "step": 9093
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7308300419976403e-07,
+      "loss": 0.1078,
+      "step": 9094
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.719813493841323e-07,
+      "loss": 0.0475,
+      "step": 9095
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7088130305894975e-07,
+      "loss": 0.0791,
+      "step": 9096
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.697828653451746e-07,
+      "loss": 0.0676,
+      "step": 9097
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.686860363635919e-07,
+      "loss": 0.038,
+      "step": 9098
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.675908162348068e-07,
+      "loss": 0.0436,
+      "step": 9099
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6649720507924976e-07,
+      "loss": 0.0372,
+      "step": 9100
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.654052030171745e-07,
+      "loss": 0.0212,
+      "step": 9101
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6431481016865664e-07,
+      "loss": 0.0826,
+      "step": 9102
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.632260266535953e-07,
+      "loss": 0.0286,
+      "step": 9103
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.62138852591713e-07,
+      "loss": 0.0702,
+      "step": 9104
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6105328810255577e-07,
+      "loss": 0.0685,
+      "step": 9105
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.599693333054882e-07,
+      "loss": 0.1054,
+      "step": 9106
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.588869883197049e-07,
+      "loss": 0.0732,
+      "step": 9107
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.57806253264219e-07,
+      "loss": 0.1094,
+      "step": 9108
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.567271282578688e-07,
+      "loss": 0.0778,
+      "step": 9109
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.5564961341931266e-07,
+      "loss": 0.0837,
+      "step": 9110
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.5457370886703424e-07,
+      "loss": 0.11,
+      "step": 9111
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.534994147193388e-07,
+      "loss": 0.0825,
+      "step": 9112
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.5242673109435686e-07,
+      "loss": 0.1417,
+      "step": 9113
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.513556581100391e-07,
+      "loss": 0.0625,
+      "step": 9114
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.502861958841613e-07,
+      "loss": 0.0609,
+      "step": 9115
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.492183445343194e-07,
+      "loss": 0.114,
+      "step": 9116
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.4815210417793617e-07,
+      "loss": 0.0708,
+      "step": 9117
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.470874749322528e-07,
+      "loss": 0.0556,
+      "step": 9118
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.4602445691433394e-07,
+      "loss": 0.0774,
+      "step": 9119
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.4496305024107446e-07,
+      "loss": 0.0947,
+      "step": 9120
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.43903255029181e-07,
+      "loss": 0.0686,
+      "step": 9121
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.428450713951886e-07,
+      "loss": 0.0724,
+      "step": 9122
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.4178849945545597e-07,
+      "loss": 0.0762,
+      "step": 9123
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.4073353932616346e-07,
+      "loss": 0.0695,
+      "step": 9124
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.3968019112331496e-07,
+      "loss": 0.1159,
+      "step": 9125
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.3862845496273274e-07,
+      "loss": 0.0842,
+      "step": 9126
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.375783309600694e-07,
+      "loss": 0.0624,
+      "step": 9127
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.3652981923079254e-07,
+      "loss": 0.0324,
+      "step": 9128
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.354829198901982e-07,
+      "loss": 0.0982,
+      "step": 9129
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.344376330534027e-07,
+      "loss": 0.0982,
+      "step": 9130
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.3339395883534406e-07,
+      "loss": 0.1155,
+      "step": 9131
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.3235189735078555e-07,
+      "loss": 0.0703,
+      "step": 9132
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.313114487143121e-07,
+      "loss": 0.1187,
+      "step": 9133
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.3027261304032883e-07,
+      "loss": 0.0296,
+      "step": 9134
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.2923539044306616e-07,
+      "loss": 0.0787,
+      "step": 9135
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.281997810365794e-07,
+      "loss": 0.1098,
+      "step": 9136
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.2716578493474256e-07,
+      "loss": 0.0862,
+      "step": 9137
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.2613340225125134e-07,
+      "loss": 0.0608,
+      "step": 9138
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.251026330996265e-07,
+      "loss": 0.0945,
+      "step": 9139
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.2407347759321246e-07,
+      "loss": 0.0915,
+      "step": 9140
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.230459358451754e-07,
+      "loss": 0.0994,
+      "step": 9141
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.220200079685015e-07,
+      "loss": 0.0817,
+      "step": 9142
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.209956940760023e-07,
+      "loss": 0.1028,
+      "step": 9143
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1997299428030923e-07,
+      "loss": 0.1209,
+      "step": 9144
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1895190869387914e-07,
+      "loss": 0.0896,
+      "step": 9145
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1793243742899216e-07,
+      "loss": 0.0865,
+      "step": 9146
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1691458059774357e-07,
+      "loss": 0.0836,
+      "step": 9147
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1589833831206395e-07,
+      "loss": 0.0768,
+      "step": 9148
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1488371068369046e-07,
+      "loss": 0.0579,
+      "step": 9149
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.13870697824199e-07,
+      "loss": 0.1077,
+      "step": 9150
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1285929984497366e-07,
+      "loss": 0.1616,
+      "step": 9151
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.118495168572305e-07,
+      "loss": 0.0581,
+      "step": 9152
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.108413489720041e-07,
+      "loss": 0.0857,
+      "step": 9153
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.098347963001541e-07,
+      "loss": 0.0592,
+      "step": 9154
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.088298589523586e-07,
+      "loss": 0.0597,
+      "step": 9155
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.078265370391192e-07,
+      "loss": 0.0609,
+      "step": 9156
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.0682483067076273e-07,
+      "loss": 0.0779,
+      "step": 9157
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.0582473995743597e-07,
+      "loss": 0.0484,
+      "step": 9158
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.0482626500910604e-07,
+      "loss": 0.0885,
+      "step": 9159
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.0382940593556843e-07,
+      "loss": 0.0837,
+      "step": 9160
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.028341628464354e-07,
+      "loss": 0.0948,
+      "step": 9161
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.0184053585114444e-07,
+      "loss": 0.073,
+      "step": 9162
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.0084852505895153e-07,
+      "loss": 0.0581,
+      "step": 9163
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9985813057894107e-07,
+      "loss": 0.0989,
+      "step": 9164
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9886935252001423e-07,
+      "loss": 0.0802,
+      "step": 9165
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9788219099089744e-07,
+      "loss": 0.0725,
+      "step": 9166
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9689664610014044e-07,
+      "loss": 0.0443,
+      "step": 9167
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9591271795610995e-07,
+      "loss": 0.049,
+      "step": 9168
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9493040666699943e-07,
+      "loss": 0.0767,
+      "step": 9169
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.939497123408258e-07,
+      "loss": 0.083,
+      "step": 9170
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.929706350854228e-07,
+      "loss": 0.056,
+      "step": 9171
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9199317500844925e-07,
+      "loss": 0.062,
+      "step": 9172
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.910173322173876e-07,
+      "loss": 0.0812,
+      "step": 9173
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.900431068195403e-07,
+      "loss": 0.0755,
+      "step": 9174
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8907049892203495e-07,
+      "loss": 0.1065,
+      "step": 9175
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.880995086318161e-07,
+      "loss": 0.0724,
+      "step": 9176
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8713013605565496e-07,
+      "loss": 0.0993,
+      "step": 9177
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.861623813001413e-07,
+      "loss": 0.065,
+      "step": 9178
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8519624447169324e-07,
+      "loss": 0.0715,
+      "step": 9179
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8423172567654253e-07,
+      "loss": 0.0524,
+      "step": 9180
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8326882502074934e-07,
+      "loss": 0.06,
+      "step": 9181
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.82307542610194e-07,
+      "loss": 0.0742,
+      "step": 9182
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8134787855057856e-07,
+      "loss": 0.0796,
+      "step": 9183
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.803898329474236e-07,
+      "loss": 0.0663,
+      "step": 9184
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.794334059060816e-07,
+      "loss": 0.0289,
+      "step": 9185
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.78478597531715e-07,
+      "loss": 0.0992,
+      "step": 9186
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.775254079293199e-07,
+      "loss": 0.1003,
+      "step": 9187
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.765738372037041e-07,
+      "loss": 0.1333,
+      "step": 9188
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.756238854595039e-07,
+      "loss": 0.1049,
+      "step": 9189
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.746755528011741e-07,
+      "loss": 0.0795,
+      "step": 9190
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.737288393329945e-07,
+      "loss": 0.0553,
+      "step": 9191
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.7278374515906534e-07,
+      "loss": 0.0536,
+      "step": 9192
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.718402703833084e-07,
+      "loss": 0.1158,
+      "step": 9193
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.7089841510946566e-07,
+      "loss": 0.0605,
+      "step": 9194
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6995817944110435e-07,
+      "loss": 0.1291,
+      "step": 9195
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6901956348161506e-07,
+      "loss": 0.0913,
+      "step": 9196
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.680825673342052e-07,
+      "loss": 0.0513,
+      "step": 9197
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.671471911019058e-07,
+      "loss": 0.0713,
+      "step": 9198
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.662134348875711e-07,
+      "loss": 0.1191,
+      "step": 9199
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.652812987938791e-07,
+      "loss": 0.1001,
+      "step": 9200
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.643507829233227e-07,
+      "loss": 0.131,
+      "step": 9201
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6342188737822184e-07,
+      "loss": 0.0876,
+      "step": 9202
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6249461226071967e-07,
+      "loss": 0.1473,
+      "step": 9203
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6156895767277975e-07,
+      "loss": 0.0721,
+      "step": 9204
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.60644923716184e-07,
+      "loss": 0.0907,
+      "step": 9205
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.597225104925377e-07,
+      "loss": 0.0829,
+      "step": 9206
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.588017181032715e-07,
+      "loss": 0.0718,
+      "step": 9207
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5788254664963764e-07,
+      "loss": 0.1298,
+      "step": 9208
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.56964996232702e-07,
+      "loss": 0.1092,
+      "step": 9209
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.560490669533622e-07,
+      "loss": 0.0681,
+      "step": 9210
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5513475891233097e-07,
+      "loss": 0.0838,
+      "step": 9211
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5422207221014793e-07,
+      "loss": 0.0805,
+      "step": 9212
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.533110069471678e-07,
+      "loss": 0.0866,
+      "step": 9213
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.524015632235738e-07,
+      "loss": 0.1271,
+      "step": 9214
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5149374113936916e-07,
+      "loss": 0.0951,
+      "step": 9215
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5058754079437587e-07,
+      "loss": 0.1192,
+      "step": 9216
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.496829622882374e-07,
+      "loss": 0.0918,
+      "step": 9217
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4878000572042423e-07,
+      "loss": 0.0715,
+      "step": 9218
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.47878671190222e-07,
+      "loss": 0.0952,
+      "step": 9219
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.469789587967447e-07,
+      "loss": 0.0797,
+      "step": 9220
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.460808686389215e-07,
+      "loss": 0.0831,
+      "step": 9221
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.451844008155052e-07,
+      "loss": 0.0911,
+      "step": 9222
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.442895554250735e-07,
+      "loss": 0.0938,
+      "step": 9223
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.433963325660227e-07,
+      "loss": 0.0995,
+      "step": 9224
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.425047323365692e-07,
+      "loss": 0.0676,
+      "step": 9225
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4161475483475636e-07,
+      "loss": 0.0697,
+      "step": 9226
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.407264001584425e-07,
+      "loss": 0.0673,
+      "step": 9227
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.398396684053128e-07,
+      "loss": 0.0493,
+      "step": 9228
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3895455967287093e-07,
+      "loss": 0.1255,
+      "step": 9229
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.380710740584424e-07,
+      "loss": 0.1243,
+      "step": 9230
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3718921165917617e-07,
+      "loss": 0.0489,
+      "step": 9231
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3630897257204132e-07,
+      "loss": 0.0587,
+      "step": 9232
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3543035689382874e-07,
+      "loss": 0.0816,
+      "step": 9233
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3455336472114952e-07,
+      "loss": 0.1014,
+      "step": 9234
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3367799615043651e-07,
+      "loss": 0.1001,
+      "step": 9235
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3280425127794768e-07,
+      "loss": 0.0565,
+      "step": 9236
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3193213019975958e-07,
+      "loss": 0.0808,
+      "step": 9237
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.310616330117671e-07,
+      "loss": 0.0668,
+      "step": 9238
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.30192759809692e-07,
+      "loss": 0.0787,
+      "step": 9239
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2932551068907458e-07,
+      "loss": 0.0792,
+      "step": 9240
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2845988574527687e-07,
+      "loss": 0.101,
+      "step": 9241
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2759588507348273e-07,
+      "loss": 0.0948,
+      "step": 9242
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2673350876869946e-07,
+      "loss": 0.0947,
+      "step": 9243
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2587275692574961e-07,
+      "loss": 0.0873,
+      "step": 9244
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2501362963928574e-07,
+      "loss": 0.0957,
+      "step": 9245
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.241561270037723e-07,
+      "loss": 0.0693,
+      "step": 9246
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2330024911350223e-07,
+      "loss": 0.0767,
+      "step": 9247
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2244599606258853e-07,
+      "loss": 0.0716,
+      "step": 9248
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2159336794496277e-07,
+      "loss": 0.1332,
+      "step": 9249
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2074236485437993e-07,
+      "loss": 0.0946,
+      "step": 9250
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.1989298688441518e-07,
+      "loss": 0.094,
+      "step": 9251
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.1904523412846712e-07,
+      "loss": 0.031,
+      "step": 9252
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.1819910667975452e-07,
+      "loss": 0.0897,
+      "step": 9253
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.1735460463131628e-07,
+      "loss": 0.0716,
+      "step": 9254
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.1651172807601482e-07,
+      "loss": 0.0874,
+      "step": 9255
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.1567047710652933e-07,
+      "loss": 0.1142,
+      "step": 9256
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.1483085181536576e-07,
+      "loss": 0.0464,
+      "step": 9257
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.1399285229485034e-07,
+      "loss": 0.0849,
+      "step": 9258
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.1315647863712595e-07,
+      "loss": 0.0896,
+      "step": 9259
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.123217309341624e-07,
+      "loss": 0.0808,
+      "step": 9260
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.1148860927774627e-07,
+      "loss": 0.1221,
+      "step": 9261
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.1065711375948758e-07,
+      "loss": 0.0885,
+      "step": 9262
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0982724447081824e-07,
+      "loss": 0.1082,
+      "step": 9263
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0899900150298855e-07,
+      "loss": 0.0747,
+      "step": 9264
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.081723849470757e-07,
+      "loss": 0.0755,
+      "step": 9265
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0734739489397026e-07,
+      "loss": 0.0964,
+      "step": 9266
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0652403143438802e-07,
+      "loss": 0.0823,
+      "step": 9267
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0570229465886493e-07,
+      "loss": 0.0741,
+      "step": 9268
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0488218465776198e-07,
+      "loss": 0.0901,
+      "step": 9269
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0406370152125708e-07,
+      "loss": 0.036,
+      "step": 9270
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.032468453393499e-07,
+      "loss": 0.0804,
+      "step": 9271
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0243161620186024e-07,
+      "loss": 0.0768,
+      "step": 9272
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.016180141984314e-07,
+      "loss": 0.0813,
+      "step": 9273
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0080603941852683e-07,
+      "loss": 0.0666,
+      "step": 9274
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9999569195143175e-07,
+      "loss": 0.0599,
+      "step": 9275
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.991869718862499e-07,
+      "loss": 0.0729,
+      "step": 9276
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.983798793119085e-07,
+      "loss": 0.1056,
+      "step": 9277
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9757441431715485e-07,
+      "loss": 0.1447,
+      "step": 9278
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.967705769905581e-07,
+      "loss": 0.0616,
+      "step": 9279
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.959683674205076e-07,
+      "loss": 0.0819,
+      "step": 9280
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9516778569521275e-07,
+      "loss": 0.1109,
+      "step": 9281
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9436883190270817e-07,
+      "loss": 0.0756,
+      "step": 9282
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9357150613084517e-07,
+      "loss": 0.1548,
+      "step": 9283
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9277580846729537e-07,
+      "loss": 0.0431,
+      "step": 9284
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9198173899955373e-07,
+      "loss": 0.1067,
+      "step": 9285
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.911892978149371e-07,
+      "loss": 0.1077,
+      "step": 9286
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.903984850005841e-07,
+      "loss": 0.1391,
+      "step": 9287
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.896093006434485e-07,
+      "loss": 0.0527,
+      "step": 9288
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8882174483030923e-07,
+      "loss": 0.082,
+      "step": 9289
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8803581764776535e-07,
+      "loss": 0.0935,
+      "step": 9290
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8725151918224103e-07,
+      "loss": 0.067,
+      "step": 9291
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8646884951997233e-07,
+      "loss": 0.0704,
+      "step": 9292
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8568780874702375e-07,
+      "loss": 0.0795,
+      "step": 9293
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8490839694927653e-07,
+      "loss": 0.0958,
+      "step": 9294
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8413061421243716e-07,
+      "loss": 0.1501,
+      "step": 9295
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.833544606220272e-07,
+      "loss": 0.1031,
+      "step": 9296
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.825799362633934e-07,
+      "loss": 0.0703,
+      "step": 9297
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8180704122170433e-07,
+      "loss": 0.0466,
+      "step": 9298
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.810357755819453e-07,
+      "loss": 0.101,
+      "step": 9299
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8026613942892177e-07,
+      "loss": 0.1031,
+      "step": 9300
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7949813284726613e-07,
+      "loss": 0.0819,
+      "step": 9301
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7873175592142743e-07,
+      "loss": 0.0825,
+      "step": 9302
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7796700873567495e-07,
+      "loss": 0.107,
+      "step": 9303
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.772038913741031e-07,
+      "loss": 0.0652,
+      "step": 9304
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.7644240392061972e-07,
+      "loss": 0.0563,
+      "step": 9305
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.7568254645895786e-07,
+      "loss": 0.1121,
+      "step": 9306
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.749243190726757e-07,
+      "loss": 0.1033,
+      "step": 9307
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.7416772184514319e-07,
+      "loss": 0.0784,
+      "step": 9308
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.7341275485955543e-07,
+      "loss": 0.0525,
+      "step": 9309
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.7265941819893272e-07,
+      "loss": 0.0813,
+      "step": 9310
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.719077119461071e-07,
+      "loss": 0.105,
+      "step": 9311
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.7115763618373747e-07,
+      "loss": 0.0833,
+      "step": 9312
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.7040919099429953e-07,
+      "loss": 0.063,
+      "step": 9313
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6966237646009574e-07,
+      "loss": 0.0648,
+      "step": 9314
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6891719266324544e-07,
+      "loss": 0.0765,
+      "step": 9315
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6817363968568645e-07,
+      "loss": 0.0234,
+      "step": 9316
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.674317176091783e-07,
+      "loss": 0.0902,
+      "step": 9317
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6669142651530578e-07,
+      "loss": 0.0961,
+      "step": 9318
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6595276648546876e-07,
+      "loss": 0.0931,
+      "step": 9319
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6521573760089225e-07,
+      "loss": 0.1145,
+      "step": 9320
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6448033994261647e-07,
+      "loss": 0.0834,
+      "step": 9321
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6374657359150836e-07,
+      "loss": 0.0921,
+      "step": 9322
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6301443862825006e-07,
+      "loss": 0.1005,
+      "step": 9323
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6228393513334882e-07,
+      "loss": 0.0379,
+      "step": 9324
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6155506318712875e-07,
+      "loss": 0.0731,
+      "step": 9325
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6082782286973907e-07,
+      "loss": 0.0959,
+      "step": 9326
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6010221426114248e-07,
+      "loss": 0.0781,
+      "step": 9327
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5937823744113013e-07,
+      "loss": 0.0887,
+      "step": 9328
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5865589248931002e-07,
+      "loss": 0.0753,
+      "step": 9329
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.579351794851086e-07,
+      "loss": 0.166,
+      "step": 9330
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5721609850777585e-07,
+      "loss": 0.0537,
+      "step": 9331
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.564986496363835e-07,
+      "loss": 0.0636,
+      "step": 9332
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5578283294982176e-07,
+      "loss": 0.0775,
+      "step": 9333
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.550686485267977e-07,
+      "loss": 0.0637,
+      "step": 9334
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5435609644584682e-07,
+      "loss": 0.074,
+      "step": 9335
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5364517678531974e-07,
+      "loss": 0.0511,
+      "step": 9336
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.52935889623389e-07,
+      "loss": 0.0788,
+      "step": 9337
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5222823503804717e-07,
+      "loss": 0.1137,
+      "step": 9338
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.51522213107107e-07,
+      "loss": 0.0571,
+      "step": 9339
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5081782390820475e-07,
+      "loss": 0.0746,
+      "step": 9340
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5011506751879013e-07,
+      "loss": 0.0689,
+      "step": 9341
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4941394401614294e-07,
+      "loss": 0.0875,
+      "step": 9342
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4871445347735658e-07,
+      "loss": 0.094,
+      "step": 9343
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4801659597934615e-07,
+      "loss": 0.0749,
+      "step": 9344
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4732037159884691e-07,
+      "loss": 0.0437,
+      "step": 9345
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.466257804124177e-07,
+      "loss": 0.1039,
+      "step": 9346
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4593282249643236e-07,
+      "loss": 0.1437,
+      "step": 9347
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4524149792709328e-07,
+      "loss": 0.0823,
+      "step": 9348
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4455180678041467e-07,
+      "loss": 0.0552,
+      "step": 9349
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4386374913223255e-07,
+      "loss": 0.076,
+      "step": 9350
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.431773250582097e-07,
+      "loss": 0.0738,
+      "step": 9351
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.424925346338224e-07,
+      "loss": 0.0816,
+      "step": 9352
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.418093779343721e-07,
+      "loss": 0.0818,
+      "step": 9353
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4112785503497706e-07,
+      "loss": 0.0608,
+      "step": 9354
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4044796601057563e-07,
+      "loss": 0.0583,
+      "step": 9355
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3976971093593138e-07,
+      "loss": 0.0798,
+      "step": 9356
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3909308988562297e-07,
+      "loss": 0.0948,
+      "step": 9357
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3841810293405088e-07,
+      "loss": 0.0964,
+      "step": 9358
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3774475015543741e-07,
+      "loss": 0.0714,
+      "step": 9359
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.37073031623825e-07,
+      "loss": 0.0951,
+      "step": 9360
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3640294741307457e-07,
+      "loss": 0.0614,
+      "step": 9361
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3573449759686552e-07,
+      "loss": 0.1022,
+      "step": 9362
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3506768224870403e-07,
+      "loss": 0.1073,
+      "step": 9363
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3440250144191313e-07,
+      "loss": 0.0557,
+      "step": 9364
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3373895524963432e-07,
+      "loss": 0.1245,
+      "step": 9365
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.330770437448292e-07,
+      "loss": 0.0911,
+      "step": 9366
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.324167670002846e-07,
+      "loss": 0.0867,
+      "step": 9367
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3175812508860074e-07,
+      "loss": 0.087,
+      "step": 9368
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3110111808220638e-07,
+      "loss": 0.0692,
+      "step": 9369
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3044574605334037e-07,
+      "loss": 0.0834,
+      "step": 9370
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2979200907407007e-07,
+      "loss": 0.0705,
+      "step": 9371
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2913990721627966e-07,
+      "loss": 0.1132,
+      "step": 9372
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2848944055167511e-07,
+      "loss": 0.086,
+      "step": 9373
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.278406091517792e-07,
+      "loss": 0.0955,
+      "step": 9374
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.271934130879382e-07,
+      "loss": 0.1096,
+      "step": 9375
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2654785243131685e-07,
+      "loss": 0.0551,
+      "step": 9376
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2590392725290168e-07,
+      "loss": 0.0874,
+      "step": 9377
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.252616376234994e-07,
+      "loss": 0.0416,
+      "step": 9378
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2462098361373353e-07,
+      "loss": 0.0841,
+      "step": 9379
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2398196529405104e-07,
+      "loss": 0.0801,
+      "step": 9380
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2334458273471904e-07,
+      "loss": 0.0828,
+      "step": 9381
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2270883600582317e-07,
+      "loss": 0.0876,
+      "step": 9382
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2207472517726914e-07,
+      "loss": 0.1109,
+      "step": 9383
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2144225031878452e-07,
+      "loss": 0.0459,
+      "step": 9384
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.208114114999137e-07,
+      "loss": 0.0726,
+      "step": 9385
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2018220879002784e-07,
+      "loss": 0.0516,
+      "step": 9386
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.195546422583099e-07,
+      "loss": 0.0812,
+      "step": 9387
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1892871197376808e-07,
+      "loss": 0.0875,
+      "step": 9388
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1830441800522895e-07,
+      "loss": 0.0802,
+      "step": 9389
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.176817604213426e-07,
+      "loss": 0.1169,
+      "step": 9390
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1706073929057093e-07,
+      "loss": 0.0786,
+      "step": 9391
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1644135468120432e-07,
+      "loss": 0.085,
+      "step": 9392
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1582360666134995e-07,
+      "loss": 0.0862,
+      "step": 9393
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1520749529893682e-07,
+      "loss": 0.1082,
+      "step": 9394
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.145930206617074e-07,
+      "loss": 0.122,
+      "step": 9395
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1398018281723266e-07,
+      "loss": 0.0828,
+      "step": 9396
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1336898183289868e-07,
+      "loss": 0.0579,
+      "step": 9397
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1275941777591503e-07,
+      "loss": 0.0844,
+      "step": 9398
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1215149071330644e-07,
+      "loss": 0.057,
+      "step": 9399
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1154520071192109e-07,
+      "loss": 0.1088,
+      "step": 9400
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1094054783842566e-07,
+      "loss": 0.077,
+      "step": 9401
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.103375321593103e-07,
+      "loss": 0.0799,
+      "step": 9402
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0973615374087865e-07,
+      "loss": 0.0616,
+      "step": 9403
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0913641264926111e-07,
+      "loss": 0.0804,
+      "step": 9404
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0853830895040328e-07,
+      "loss": 0.0819,
+      "step": 9405
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0794184271007257e-07,
+      "loss": 0.0727,
+      "step": 9406
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.073470139938565e-07,
+      "loss": 0.0649,
+      "step": 9407
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0675382286716106e-07,
+      "loss": 0.0774,
+      "step": 9408
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0616226939521578e-07,
+      "loss": 0.0899,
+      "step": 9409
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0557235364306528e-07,
+      "loss": 0.0804,
+      "step": 9410
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0498407567557767e-07,
+      "loss": 0.0796,
+      "step": 9411
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0439743555743952e-07,
+      "loss": 0.1737,
+      "step": 9412
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0381243335315593e-07,
+      "loss": 0.1237,
+      "step": 9413
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0322906912705543e-07,
+      "loss": 0.0387,
+      "step": 9414
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0264734294328337e-07,
+      "loss": 0.0604,
+      "step": 9415
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0206725486580859e-07,
+      "loss": 0.1052,
+      "step": 9416
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0148880495841339e-07,
+      "loss": 0.0994,
+      "step": 9417
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0091199328470691e-07,
+      "loss": 0.0636,
+      "step": 9418
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0033681990811339e-07,
+      "loss": 0.0924,
+      "step": 9419
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.976328489187891e-08,
+      "loss": 0.0746,
+      "step": 9420
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.919138829906971e-08,
+      "loss": 0.0793,
+      "step": 9421
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.862113019257214e-08,
+      "loss": 0.0659,
+      "step": 9422
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.805251063508935e-08,
+      "loss": 0.0579,
+      "step": 9423
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.748552968914803e-08,
+      "loss": 0.0845,
+      "step": 9424
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.692018741709164e-08,
+      "loss": 0.115,
+      "step": 9425
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.635648388108708e-08,
+      "loss": 0.0789,
+      "step": 9426
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.579441914311816e-08,
+      "loss": 0.0592,
+      "step": 9427
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.523399326498872e-08,
+      "loss": 0.0717,
+      "step": 9428
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.467520630832282e-08,
+      "loss": 0.0806,
+      "step": 9429
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.411805833456465e-08,
+      "loss": 0.0775,
+      "step": 9430
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.356254940498021e-08,
+      "loss": 0.0923,
+      "step": 9431
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.300867958064896e-08,
+      "loss": 0.0445,
+      "step": 9432
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.245644892247718e-08,
+      "loss": 0.0263,
+      "step": 9433
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.190585749118463e-08,
+      "loss": 0.0951,
+      "step": 9434
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.135690534731955e-08,
+      "loss": 0.114,
+      "step": 9435
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.080959255123866e-08,
+      "loss": 0.0929,
+      "step": 9436
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.026391916312882e-08,
+      "loss": 0.1109,
+      "step": 9437
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.971988524298869e-08,
+      "loss": 0.0626,
+      "step": 9438
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.917749085064208e-08,
+      "loss": 0.0962,
+      "step": 9439
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.863673604572964e-08,
+      "loss": 0.1172,
+      "step": 9440
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.809762088771211e-08,
+      "loss": 0.0656,
+      "step": 9441
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.75601454358721e-08,
+      "loss": 0.0561,
+      "step": 9442
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.702430974930731e-08,
+      "loss": 0.058,
+      "step": 9443
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.649011388694228e-08,
+      "loss": 0.0574,
+      "step": 9444
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.59575579075117e-08,
+      "loss": 0.0965,
+      "step": 9445
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.542664186957871e-08,
+      "loss": 0.097,
+      "step": 9446
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.489736583151997e-08,
+      "loss": 0.1049,
+      "step": 9447
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.43697298515389e-08,
+      "loss": 0.059,
+      "step": 9448
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.384373398765077e-08,
+      "loss": 0.08,
+      "step": 9449
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.331937829769431e-08,
+      "loss": 0.092,
+      "step": 9450
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.279666283932674e-08,
+      "loss": 0.0832,
+      "step": 9451
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.227558767002707e-08,
+      "loss": 0.1194,
+      "step": 9452
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.175615284709114e-08,
+      "loss": 0.065,
+      "step": 9453
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.123835842763827e-08,
+      "loss": 0.0744,
+      "step": 9454
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.072220446860123e-08,
+      "loss": 0.0457,
+      "step": 9455
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.020769102673964e-08,
+      "loss": 0.0498,
+      "step": 9456
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.969481815862656e-08,
+      "loss": 0.0417,
+      "step": 9457
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.918358592065689e-08,
+      "loss": 0.0671,
+      "step": 9458
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.867399436904898e-08,
+      "loss": 0.1132,
+      "step": 9459
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.81660435598347e-08,
+      "loss": 0.103,
+      "step": 9460
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.765973354886768e-08,
+      "loss": 0.0674,
+      "step": 9461
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.71550643918234e-08,
+      "loss": 0.0718,
+      "step": 9462
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.665203614419414e-08,
+      "loss": 0.1197,
+      "step": 9463
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.615064886129231e-08,
+      "loss": 0.0828,
+      "step": 9464
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.565090259825047e-08,
+      "loss": 0.1082,
+      "step": 9465
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.515279741002135e-08,
+      "loss": 0.0696,
+      "step": 9466
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.465633335137445e-08,
+      "loss": 0.0465,
+      "step": 9467
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.416151047690112e-08,
+      "loss": 0.0718,
+      "step": 9468
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.366832884101448e-08,
+      "loss": 0.0566,
+      "step": 9469
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.317678849794119e-08,
+      "loss": 0.0648,
+      "step": 9470
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.268688950173297e-08,
+      "loss": 0.0488,
+      "step": 9471
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.219863190625841e-08,
+      "loss": 0.1035,
+      "step": 9472
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.171201576520626e-08,
+      "loss": 0.0833,
+      "step": 9473
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.122704113208367e-08,
+      "loss": 0.0703,
+      "step": 9474
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.074370806021802e-08,
+      "loss": 0.0581,
+      "step": 9475
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.026201660275844e-08,
+      "loss": 0.1192,
+      "step": 9476
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.97819668126709e-08,
+      "loss": 0.1042,
+      "step": 9477
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.930355874273985e-08,
+      "loss": 0.0685,
+      "step": 9478
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.882679244557156e-08,
+      "loss": 0.0439,
+      "step": 9479
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.835166797359237e-08,
+      "loss": 0.0618,
+      "step": 9480
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.787818537904555e-08,
+      "loss": 0.1309,
+      "step": 9481
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.740634471399442e-08,
+      "loss": 0.1114,
+      "step": 9482
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.69361460303225e-08,
+      "loss": 0.1125,
+      "step": 9483
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.646758937973508e-08,
+      "loss": 0.0805,
+      "step": 9484
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.600067481375094e-08,
+      "loss": 0.0725,
+      "step": 9485
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.553540238371402e-08,
+      "loss": 0.1353,
+      "step": 9486
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.507177214078508e-08,
+      "loss": 0.0497,
+      "step": 9487
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.460978413594498e-08,
+      "loss": 0.0691,
+      "step": 9488
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.414943841999476e-08,
+      "loss": 0.0695,
+      "step": 9489
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.369073504355061e-08,
+      "loss": 0.0786,
+      "step": 9490
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.323367405705216e-08,
+      "loss": 0.0772,
+      "step": 9491
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.277825551075922e-08,
+      "loss": 0.0679,
+      "step": 9492
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.232447945475006e-08,
+      "loss": 0.0574,
+      "step": 9493
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.187234593891978e-08,
+      "loss": 0.0621,
+      "step": 9494
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.142185501298358e-08,
+      "loss": 0.0629,
+      "step": 9495
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.097300672648021e-08,
+      "loss": 0.1243,
+      "step": 9496
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.052580112876516e-08,
+      "loss": 0.0396,
+      "step": 9497
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.008023826901076e-08,
+      "loss": 0.1061,
+      "step": 9498
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.963631819621118e-08,
+      "loss": 0.0684,
+      "step": 9499
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.919404095917902e-08,
+      "loss": 0.0817,
+      "step": 9500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8753406606550374e-08,
+      "loss": 0.0911,
+      "step": 9501
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8314415186773166e-08,
+      "loss": 0.085,
+      "step": 9502
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7877066748120434e-08,
+      "loss": 0.1349,
+      "step": 9503
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.744136133868372e-08,
+      "loss": 0.0473,
+      "step": 9504
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7007299006371364e-08,
+      "loss": 0.0496,
+      "step": 9505
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.6574879798913515e-08,
+      "loss": 0.1137,
+      "step": 9506
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.6144103763857144e-08,
+      "loss": 0.1137,
+      "step": 9507
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.57149709485727e-08,
+      "loss": 0.0792,
+      "step": 9508
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.528748140024742e-08,
+      "loss": 0.0867,
+      "step": 9509
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.4861635165887046e-08,
+      "loss": 0.1235,
+      "step": 9510
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.4437432292315793e-08,
+      "loss": 0.1241,
+      "step": 9511
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.4014872826179674e-08,
+      "loss": 0.074,
+      "step": 9512
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.359395681394486e-08,
+      "loss": 0.0887,
+      "step": 9513
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.317468430189432e-08,
+      "loss": 0.0832,
+      "step": 9514
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.275705533612951e-08,
+      "loss": 0.0331,
+      "step": 9515
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.234106996257537e-08,
+      "loss": 0.0662,
+      "step": 9516
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.192672822697031e-08,
+      "loss": 0.0583,
+      "step": 9517
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.1514030174877877e-08,
+      "loss": 0.0771,
+      "step": 9518
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.1102975851678446e-08,
+      "loss": 0.0693,
+      "step": 9519
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.069356530256752e-08,
+      "loss": 0.0532,
+      "step": 9520
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.028579857256743e-08,
+      "loss": 0.0679,
+      "step": 9521
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.987967570651564e-08,
+      "loss": 0.0669,
+      "step": 9522
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.947519674906809e-08,
+      "loss": 0.0432,
+      "step": 9523
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.907236174470087e-08,
+      "loss": 0.0988,
+      "step": 9524
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.867117073771021e-08,
+      "loss": 0.0847,
+      "step": 9525
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.827162377221084e-08,
+      "loss": 0.0677,
+      "step": 9526
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.7873720892137596e-08,
+      "loss": 0.0713,
+      "step": 9527
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.747746214124049e-08,
+      "loss": 0.0921,
+      "step": 9528
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.708284756309633e-08,
+      "loss": 0.0785,
+      "step": 9529
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.668987720109374e-08,
+      "loss": 0.057,
+      "step": 9530
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.629855109844649e-08,
+      "loss": 0.0801,
+      "step": 9531
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.5908869298181835e-08,
+      "loss": 0.1389,
+      "step": 9532
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.5520831843148815e-08,
+      "loss": 0.0806,
+      "step": 9533
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.513443877601831e-08,
+      "loss": 0.062,
+      "step": 9534
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.4749690139276325e-08,
+      "loss": 0.1351,
+      "step": 9535
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.436658597522902e-08,
+      "loss": 0.0729,
+      "step": 9536
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.398512632600271e-08,
+      "loss": 0.0581,
+      "step": 9537
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.3605311233545495e-08,
+      "loss": 0.084,
+      "step": 9538
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.322714073961731e-08,
+      "loss": 0.1044,
+      "step": 9539
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.285061488580322e-08,
+      "loss": 0.1352,
+      "step": 9540
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.247573371350677e-08,
+      "loss": 0.0833,
+      "step": 9541
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.210249726394999e-08,
+      "loss": 0.0392,
+      "step": 9542
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.173090557817172e-08,
+      "loss": 0.1182,
+      "step": 9543
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.136095869703427e-08,
+      "loss": 0.094,
+      "step": 9544
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.09926566612151e-08,
+      "loss": 0.0608,
+      "step": 9545
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0625999511213486e-08,
+      "loss": 0.1027,
+      "step": 9546
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.026098728734717e-08,
+      "loss": 0.1198,
+      "step": 9547
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.9897620029750724e-08,
+      "loss": 0.0583,
+      "step": 9548
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.953589777838218e-08,
+      "loss": 0.1217,
+      "step": 9549
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.917582057301472e-08,
+      "loss": 0.0472,
+      "step": 9550
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.881738845324501e-08,
+      "loss": 0.0917,
+      "step": 9551
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.8460601458481515e-08,
+      "loss": 0.0681,
+      "step": 9552
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.810545962795953e-08,
+      "loss": 0.0792,
+      "step": 9553
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.775196300073114e-08,
+      "loss": 0.0856,
+      "step": 9554
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.740011161566526e-08,
+      "loss": 0.1216,
+      "step": 9555
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.704990551145093e-08,
+      "loss": 0.0742,
+      "step": 9556
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.67013447265957e-08,
+      "loss": 0.0486,
+      "step": 9557
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.635442929942889e-08,
+      "loss": 0.1105,
+      "step": 9558
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.600915926809834e-08,
+      "loss": 0.0569,
+      "step": 9559
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.5665534670566994e-08,
+      "loss": 0.1247,
+      "step": 9560
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.53235555446213e-08,
+      "loss": 0.053,
+      "step": 9561
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.498322192786452e-08,
+      "loss": 0.1074,
+      "step": 9562
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.464453385772004e-08,
+      "loss": 0.0747,
+      "step": 9563
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.430749137142974e-08,
+      "loss": 0.0891,
+      "step": 9564
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.397209450605398e-08,
+      "loss": 0.0541,
+      "step": 9565
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.363834329847326e-08,
+      "loss": 0.1044,
+      "step": 9566
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.330623778538822e-08,
+      "loss": 0.0539,
+      "step": 9567
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.297577800331464e-08,
+      "loss": 0.0722,
+      "step": 9568
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.264696398859013e-08,
+      "loss": 0.1527,
+      "step": 9569
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.2319795777372427e-08,
+      "loss": 0.073,
+      "step": 9570
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.19942734056361e-08,
+      "loss": 0.0807,
+      "step": 9571
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.1670396909175836e-08,
+      "loss": 0.0599,
+      "step": 9572
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.134816632360316e-08,
+      "loss": 0.0639,
+      "step": 9573
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.1027581684353044e-08,
+      "loss": 0.0751,
+      "step": 9574
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0708643026673977e-08,
+      "loss": 0.0627,
+      "step": 9575
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.03913503856379e-08,
+      "loss": 0.1196,
+      "step": 9576
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0075703796135226e-08,
+      "loss": 0.0489,
+      "step": 9577
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.9761703292871535e-08,
+      "loss": 0.0906,
+      "step": 9578
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.9449348910377538e-08,
+      "loss": 0.0925,
+      "step": 9579
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.9138640682995764e-08,
+      "loss": 0.0601,
+      "step": 9580
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8829578644895547e-08,
+      "loss": 0.1102,
+      "step": 9581
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.852216283005804e-08,
+      "loss": 0.0652,
+      "step": 9582
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8216393272287865e-08,
+      "loss": 0.1024,
+      "step": 9583
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.791227000520813e-08,
+      "loss": 0.0816,
+      "step": 9584
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.760979306225875e-08,
+      "loss": 0.0646,
+      "step": 9585
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.7308962476699784e-08,
+      "loss": 0.0729,
+      "step": 9586
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.7009778281611443e-08,
+      "loss": 0.0751,
+      "step": 9587
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.6712240509892405e-08,
+      "loss": 0.0821,
+      "step": 9588
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.641634919425817e-08,
+      "loss": 0.0635,
+      "step": 9589
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.6122104367246047e-08,
+      "loss": 0.0986,
+      "step": 9590
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.582950606121015e-08,
+      "loss": 0.0225,
+      "step": 9591
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.553855430832308e-08,
+      "loss": 0.0976,
+      "step": 9592
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5249249140582574e-08,
+      "loss": 0.0451,
+      "step": 9593
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4961590589796523e-08,
+      "loss": 0.1143,
+      "step": 9594
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4675578687594625e-08,
+      "loss": 0.1291,
+      "step": 9595
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.439121346543005e-08,
+      "loss": 0.1308,
+      "step": 9596
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4108494954571124e-08,
+      "loss": 0.0851,
+      "step": 9597
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.382742318610298e-08,
+      "loss": 0.0593,
+      "step": 9598
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.354799819093423e-08,
+      "loss": 0.0933,
+      "step": 9599
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3270219999790286e-08,
+      "loss": 0.1143,
+      "step": 9600
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.299408864321506e-08,
+      "loss": 0.062,
+      "step": 9601
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.271960415156926e-08,
+      "loss": 0.0895,
+      "step": 9602
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.244676655504041e-08,
+      "loss": 0.0846,
+      "step": 9603
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.217557588362451e-08,
+      "loss": 0.0733,
+      "step": 9604
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.1906032167144374e-08,
+      "loss": 0.1185,
+      "step": 9605
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.1638135435239626e-08,
+      "loss": 0.088,
+      "step": 9606
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.1371885717365035e-08,
+      "loss": 0.0599,
+      "step": 9607
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.1107283042798854e-08,
+      "loss": 0.0545,
+      "step": 9608
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.08443274406378e-08,
+      "loss": 0.0492,
+      "step": 9609
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.058301893979542e-08,
+      "loss": 0.1709,
+      "step": 9610
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.0323357569005385e-08,
+      "loss": 0.0587,
+      "step": 9611
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.00653433568182e-08,
+      "loss": 0.0626,
+      "step": 9612
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.980897633160783e-08,
+      "loss": 0.0708,
+      "step": 9613
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.955425652156173e-08,
+      "loss": 0.1203,
+      "step": 9614
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9301183954690826e-08,
+      "loss": 0.0821,
+      "step": 9615
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.904975865882119e-08,
+      "loss": 0.058,
+      "step": 9616
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8799980661600713e-08,
+      "loss": 0.0765,
+      "step": 9617
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8551849990494085e-08,
+      "loss": 0.0691,
+      "step": 9618
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8305366672786152e-08,
+      "loss": 0.0746,
+      "step": 9619
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8060530735580227e-08,
+      "loss": 0.1349,
+      "step": 9620
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7817342205796448e-08,
+      "loss": 0.1296,
+      "step": 9621
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7575801110178424e-08,
+      "loss": 0.0967,
+      "step": 9622
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7335907475284908e-08,
+      "loss": 0.2927,
+      "step": 9623
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.709766132749313e-08,
+      "loss": 0.0924,
+      "step": 9624
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6861062693002138e-08,
+      "loss": 0.0825,
+      "step": 9625
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6626111597829454e-08,
+      "loss": 0.0952,
+      "step": 9626
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6392808067807742e-08,
+      "loss": 0.0565,
+      "step": 9627
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.616115212858982e-08,
+      "loss": 0.0536,
+      "step": 9628
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.593114380565197e-08,
+      "loss": 0.0457,
+      "step": 9629
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5702783124285636e-08,
+      "loss": 0.0446,
+      "step": 9630
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5476070109597396e-08,
+      "loss": 0.0823,
+      "step": 9631
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.525100478652064e-08,
+      "loss": 0.045,
+      "step": 9632
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5027587179800572e-08,
+      "loss": 0.0905,
+      "step": 9633
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4805817314005877e-08,
+      "loss": 0.0961,
+      "step": 9634
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4585695213522043e-08,
+      "loss": 0.0952,
+      "step": 9635
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4367220902553046e-08,
+      "loss": 0.0901,
+      "step": 9636
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4150394405123002e-08,
+      "loss": 0.0277,
+      "step": 9637
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3935215745072838e-08,
+      "loss": 0.0726,
+      "step": 9638
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3721684946063628e-08,
+      "loss": 0.0697,
+      "step": 9639
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.350980203157659e-08,
+      "loss": 0.0998,
+      "step": 9640
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3299567024908088e-08,
+      "loss": 0.0597,
+      "step": 9641
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3090979949177962e-08,
+      "loss": 0.1242,
+      "step": 9642
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2884040827321198e-08,
+      "loss": 0.0909,
+      "step": 9643
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2678749682092928e-08,
+      "loss": 0.0794,
+      "step": 9644
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2475106536065095e-08,
+      "loss": 0.039,
+      "step": 9645
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.227311141163312e-08,
+      "loss": 0.0563,
+      "step": 9646
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2072764331005903e-08,
+      "loss": 0.0631,
+      "step": 9647
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1874065316214155e-08,
+      "loss": 0.1425,
+      "step": 9648
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1677014389108732e-08,
+      "loss": 0.0834,
+      "step": 9649
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1481611571353968e-08,
+      "loss": 0.0897,
+      "step": 9650
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.128785688443934e-08,
+      "loss": 0.0858,
+      "step": 9651
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.109575034966781e-08,
+      "loss": 0.0823,
+      "step": 9652
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0905291988164145e-08,
+      "loss": 0.0938,
+      "step": 9653
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0716481820871593e-08,
+      "loss": 0.1417,
+      "step": 9654
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0529319868550213e-08,
+      "loss": 0.0896,
+      "step": 9655
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0343806151781876e-08,
+      "loss": 0.0558,
+      "step": 9656
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0159940690965264e-08,
+      "loss": 0.074,
+      "step": 9657
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.977723506315873e-09,
+      "loss": 0.09,
+      "step": 9658
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.797154617874337e-09,
+      "loss": 0.1495,
+      "step": 9659
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.618234045493113e-09,
+      "loss": 0.1021,
+      "step": 9660
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.440961808844795e-09,
+      "loss": 0.0449,
+      "step": 9661
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.265337927425455e-09,
+      "loss": 0.0771,
+      "step": 9662
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.091362420546313e-09,
+      "loss": 0.077,
+      "step": 9663
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.9190353073354e-09,
+      "loss": 0.0818,
+      "step": 9664
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.748356606742559e-09,
+      "loss": 0.0761,
+      "step": 9665
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.579326337537774e-09,
+      "loss": 0.054,
+      "step": 9666
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.411944518304516e-09,
+      "loss": 0.0718,
+      "step": 9667
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.246211167449724e-09,
+      "loss": 0.1287,
+      "step": 9668
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.082126303195491e-09,
+      "loss": 0.0822,
+      "step": 9669
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.919689943587382e-09,
+      "loss": 0.0403,
+      "step": 9670
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.758902106484445e-09,
+      "loss": 0.0659,
+      "step": 9671
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.5997628095692e-09,
+      "loss": 0.1159,
+      "step": 9672
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.442272070337652e-09,
+      "loss": 0.0743,
+      "step": 9673
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.286429906107617e-09,
+      "loss": 0.0597,
+      "step": 9674
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.132236334018716e-09,
+      "loss": 0.0722,
+      "step": 9675
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.9796913710223895e-09,
+      "loss": 0.0567,
+      "step": 9676
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.828795033895219e-09,
+      "loss": 0.0591,
+      "step": 9677
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.679547339228931e-09,
+      "loss": 0.1108,
+      "step": 9678
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.531948303432067e-09,
+      "loss": 0.0969,
+      "step": 9679
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.385997942738308e-09,
+      "loss": 0.0964,
+      "step": 9680
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.241696273194819e-09,
+      "loss": 0.083,
+      "step": 9681
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.09904331066724e-09,
+      "loss": 0.0634,
+      "step": 9682
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.9580390708446896e-09,
+      "loss": 0.0481,
+      "step": 9683
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.818683569229765e-09,
+      "loss": 0.0632,
+      "step": 9684
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.680976821146877e-09,
+      "loss": 0.0642,
+      "step": 9685
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.544918841738911e-09,
+      "loss": 0.0797,
+      "step": 9686
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.410509645963902e-09,
+      "loss": 0.124,
+      "step": 9687
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.277749248605024e-09,
+      "loss": 0.0966,
+      "step": 9688
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.14663766425727e-09,
+      "loss": 0.0985,
+      "step": 9689
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.017174907339106e-09,
+      "loss": 0.0944,
+      "step": 9690
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.889360992087477e-09,
+      "loss": 0.0379,
+      "step": 9691
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.763195932554476e-09,
+      "loss": 0.0577,
+      "step": 9692
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.63867974261567e-09,
+      "loss": 0.0793,
+      "step": 9693
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.515812435960109e-09,
+      "loss": 0.0883,
+      "step": 9694
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.394594026100318e-09,
+      "loss": 0.0856,
+      "step": 9695
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.275024526363969e-09,
+      "loss": 0.081,
+      "step": 9696
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.157103949898877e-09,
+      "loss": 0.0772,
+      "step": 9697
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.040832309673004e-09,
+      "loss": 0.0955,
+      "step": 9698
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.926209618469456e-09,
+      "loss": 0.0399,
+      "step": 9699
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.813235888894817e-09,
+      "loss": 0.0869,
+      "step": 9700
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7019111333691513e-09,
+      "loss": 0.0783,
+      "step": 9701
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.5922353641359986e-09,
+      "loss": 0.0446,
+      "step": 9702
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4842085932523803e-09,
+      "loss": 0.0921,
+      "step": 9703
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.3778308325987937e-09,
+      "loss": 0.1382,
+      "step": 9704
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.273102093872549e-09,
+      "loss": 0.115,
+      "step": 9705
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.1700223885877676e-09,
+      "loss": 0.0322,
+      "step": 9706
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0685917280820487e-09,
+      "loss": 0.079,
+      "step": 9707
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.968810123504806e-09,
+      "loss": 0.0639,
+      "step": 9708
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.8706775858322597e-09,
+      "loss": 0.0403,
+      "step": 9709
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7741941258524472e-09,
+      "loss": 0.1344,
+      "step": 9710
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.6793597541735494e-09,
+      "loss": 0.1207,
+      "step": 9711
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5861744812272214e-09,
+      "loss": 0.0771,
+      "step": 9712
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.4946383172569367e-09,
+      "loss": 0.0824,
+      "step": 9713
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.4047512723296418e-09,
+      "loss": 0.1166,
+      "step": 9714
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.316513356327432e-09,
+      "loss": 0.0616,
+      "step": 9715
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.2299245789542123e-09,
+      "loss": 0.1164,
+      "step": 9716
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.144984949732365e-09,
+      "loss": 0.066,
+      "step": 9717
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.0616944779994205e-09,
+      "loss": 0.0804,
+      "step": 9718
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.980053172916385e-09,
+      "loss": 0.1069,
+      "step": 9719
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.900061043457746e-09,
+      "loss": 0.063,
+      "step": 9720
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8217180984214654e-09,
+      "loss": 0.0689,
+      "step": 9721
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7450243464223191e-09,
+      "loss": 0.1383,
+      "step": 9722
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6699797958902307e-09,
+      "loss": 0.0459,
+      "step": 9723
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5965844550819287e-09,
+      "loss": 0.074,
+      "step": 9724
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5248383320642934e-09,
+      "loss": 0.0712,
+      "step": 9725
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.4547414347293452e-09,
+      "loss": 0.0363,
+      "step": 9726
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3862937707825873e-09,
+      "loss": 0.0944,
+      "step": 9727
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3194953477496663e-09,
+      "loss": 0.0691,
+      "step": 9728
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2543461729797035e-09,
+      "loss": 0.0867,
+      "step": 9729
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.190846253631972e-09,
+      "loss": 0.0576,
+      "step": 9730
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1289955966925503e-09,
+      "loss": 0.0449,
+      "step": 9731
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.068794208959334e-09,
+      "loss": 0.094,
+      "step": 9732
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0102420970553583e-09,
+      "loss": 0.0572,
+      "step": 9733
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.533392674154762e-10,
+      "loss": 0.1406,
+      "step": 9734
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.9808572629968e-10,
+      "loss": 0.0825,
+      "step": 9735
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.444814797814449e-10,
+      "loss": 0.0969,
+      "step": 9736
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.925265337577203e-10,
+      "loss": 0.0887,
+      "step": 9737
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.422208939389385e-10,
+      "loss": 0.0879,
+      "step": 9738
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.935645658573409e-10,
+      "loss": 0.1281,
+      "step": 9739
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.465575548619818e-10,
+      "loss": 0.084,
+      "step": 9740
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.011998661253904e-10,
+      "loss": 0.0711,
+      "step": 9741
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.574915046319129e-10,
+      "loss": 0.0688,
+      "step": 9742
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.154324751877049e-10,
+      "loss": 0.0536,
+      "step": 9743
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.750227824190656e-10,
+      "loss": 0.0744,
+      "step": 9744
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.3626243076910764e-10,
+      "loss": 0.1343,
+      "step": 9745
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.991514244994221e-10,
+      "loss": 0.0826,
+      "step": 9746
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.6368976769007857e-10,
+      "loss": 0.1686,
+      "step": 9747
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.2987746424295586e-10,
+      "loss": 0.064,
+      "step": 9748
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.9771451787174997e-10,
+      "loss": 0.0563,
+      "step": 9749
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.672009321186275e-10,
+      "loss": 0.0874,
+      "step": 9750
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3833671033424156e-10,
+      "loss": 0.0755,
+      "step": 9751
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.111218556927197e-10,
+      "loss": 0.0926,
+      "step": 9752
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.855563711899988e-10,
+      "loss": 0.0662,
+      "step": 9753
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6164025963549823e-10,
+      "loss": 0.112,
+      "step": 9754
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3937352365711588e-10,
+      "loss": 0.0686,
+      "step": 9755
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1875616570622415e-10,
+      "loss": 0.0956,
+      "step": 9756
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.978818804767808e-11,
+      "loss": 0.0901,
+      "step": 9757
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.246959276814181e-11,
+      "loss": 0.0831,
+      "step": 9758
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.680038177275805e-11,
+      "loss": 0.0719,
+      "step": 9759
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2780556783482705e-11,
+      "loss": 0.1007,
+      "step": 9760
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0410119340750226e-11,
+      "loss": 0.0391,
+      "step": 9761
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.968907080846961e-11,
+      "loss": 0.0841,
+      "step": 9762
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0617412360701694e-11,
+      "loss": 0.0578,
+      "step": 9763
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.319514499997787e-11,
+      "loss": 0.0745,
+      "step": 9764
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.422269538981397e-12,
+      "loss": 0.0732,
+      "step": 9765
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.2987866122047296e-12,
+      "loss": 0.0755,
+      "step": 9766
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.246966759495323e-13,
+      "loss": 0.0994,
+      "step": 9767
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0,
+      "loss": 0.0796,
+      "step": 9768
+    },
+    {
+      "epoch": 2.0,
+      "step": 9768,
+      "total_flos": 367896721858560.0,
+      "train_loss": 0.20280386607584458,
+      "train_runtime": 96935.5475,
+      "train_samples_per_second": 2.418,
+      "train_steps_per_second": 0.101
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 9768,
+  "num_train_epochs": 2,
+  "save_steps": 50000,
+  "total_flos": 367896721858560.0,
+  "trial_name": null,
+  "trial_params": null
+}