diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,29329 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "global_step": 4884,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0408163265306121e-07,
+      "loss": 1.5625,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0816326530612243e-07,
+      "loss": 1.3834,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.122448979591837e-07,
+      "loss": 1.512,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.163265306122449e-07,
+      "loss": 1.6229,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0204081632653063e-06,
+      "loss": 1.4121,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2244897959183673e-06,
+      "loss": 1.5229,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4285714285714286e-06,
+      "loss": 1.5299,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6326530612244897e-06,
+      "loss": 1.3765,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.836734693877551e-06,
+      "loss": 1.4,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0408163265306125e-06,
+      "loss": 1.4369,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2448979591836734e-06,
+      "loss": 1.2036,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4489795918367347e-06,
+      "loss": 1.2208,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.653061224489796e-06,
+      "loss": 1.1757,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 1.1769,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0612244897959185e-06,
+      "loss": 1.0721,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2653061224489794e-06,
+      "loss": 1.207,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.469387755102041e-06,
+      "loss": 1.1056,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.673469387755102e-06,
+      "loss": 1.0788,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.877551020408164e-06,
+      "loss": 1.1398,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.081632653061225e-06,
+      "loss": 0.9997,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 1.0639,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.489795918367347e-06,
+      "loss": 1.0198,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.693877551020409e-06,
+      "loss": 0.9486,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.897959183673469e-06,
+      "loss": 0.9178,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.102040816326531e-06,
+      "loss": 0.9109,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.306122448979592e-06,
+      "loss": 0.8938,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.510204081632653e-06,
+      "loss": 0.9198,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 0.8834,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.918367346938776e-06,
+      "loss": 0.9241,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.122448979591837e-06,
+      "loss": 0.935,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.326530612244898e-06,
+      "loss": 0.8717,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.530612244897959e-06,
+      "loss": 0.8456,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.734693877551021e-06,
+      "loss": 0.8331,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.938775510204082e-06,
+      "loss": 0.9132,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.142857142857143e-06,
+      "loss": 0.8719,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.346938775510204e-06,
+      "loss": 0.9055,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.551020408163265e-06,
+      "loss": 0.8584,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.755102040816327e-06,
+      "loss": 0.8151,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.959183673469388e-06,
+      "loss": 0.7769,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.16326530612245e-06,
+      "loss": 0.858,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.36734693877551e-06,
+      "loss": 0.8771,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.8018,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.775510204081633e-06,
+      "loss": 0.7994,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.979591836734694e-06,
+      "loss": 0.7477,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.183673469387756e-06,
+      "loss": 0.7833,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.387755102040818e-06,
+      "loss": 0.8282,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.591836734693878e-06,
+      "loss": 0.7972,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.795918367346939e-06,
+      "loss": 0.7595,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999999999999999e-06,
+      "loss": 0.7443,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0204081632653061e-05,
+      "loss": 0.7496,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0408163265306123e-05,
+      "loss": 0.7414,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0612244897959184e-05,
+      "loss": 0.7297,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0816326530612246e-05,
+      "loss": 0.7341,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1020408163265306e-05,
+      "loss": 0.7737,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1224489795918369e-05,
+      "loss": 0.7767,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 0.7522,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.163265306122449e-05,
+      "loss": 0.7559,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1836734693877552e-05,
+      "loss": 0.8293,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2040816326530612e-05,
+      "loss": 0.7533,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2244897959183674e-05,
+      "loss": 0.656,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2448979591836736e-05,
+      "loss": 0.7022,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2653061224489797e-05,
+      "loss": 0.7219,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2857142857142857e-05,
+      "loss": 0.7224,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3061224489795918e-05,
+      "loss": 0.7338,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.326530612244898e-05,
+      "loss": 0.6905,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3469387755102042e-05,
+      "loss": 0.7441,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3673469387755102e-05,
+      "loss": 0.7008,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3877551020408165e-05,
+      "loss": 0.7712,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4081632653061225e-05,
+      "loss": 0.7379,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 0.7355,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4489795918367348e-05,
+      "loss": 0.6917,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4693877551020408e-05,
+      "loss": 0.7068,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.489795918367347e-05,
+      "loss": 0.7005,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.510204081632653e-05,
+      "loss": 0.659,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.530612244897959e-05,
+      "loss": 0.7125,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5510204081632655e-05,
+      "loss": 0.6185,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5714285714285715e-05,
+      "loss": 0.6979,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5918367346938776e-05,
+      "loss": 0.6853,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.612244897959184e-05,
+      "loss": 0.6641,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.63265306122449e-05,
+      "loss": 0.7072,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6530612244897957e-05,
+      "loss": 0.7106,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.673469387755102e-05,
+      "loss": 0.6325,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.693877551020408e-05,
+      "loss": 0.703,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 0.615,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7346938775510202e-05,
+      "loss": 0.7312,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7551020408163266e-05,
+      "loss": 0.6054,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7755102040816327e-05,
+      "loss": 0.7084,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7959183673469387e-05,
+      "loss": 0.6413,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.816326530612245e-05,
+      "loss": 0.6418,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.836734693877551e-05,
+      "loss": 0.6915,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8571428571428572e-05,
+      "loss": 0.6766,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8775510204081636e-05,
+      "loss": 0.7174,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8979591836734696e-05,
+      "loss": 0.6343,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9183673469387756e-05,
+      "loss": 0.6008,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9387755102040817e-05,
+      "loss": 0.7268,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9591836734693877e-05,
+      "loss": 0.5988,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9795918367346938e-05,
+      "loss": 0.7244,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9999999999999998e-05,
+      "loss": 0.6618,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0204081632653062e-05,
+      "loss": 0.7073,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0408163265306123e-05,
+      "loss": 0.7082,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0612244897959183e-05,
+      "loss": 0.6559,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0816326530612247e-05,
+      "loss": 0.6046,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1020408163265307e-05,
+      "loss": 0.6759,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1224489795918368e-05,
+      "loss": 0.6416,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1428571428571428e-05,
+      "loss": 0.6022,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1632653061224492e-05,
+      "loss": 0.6709,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1836734693877552e-05,
+      "loss": 0.6762,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2040816326530613e-05,
+      "loss": 0.7006,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2244897959183677e-05,
+      "loss": 0.69,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2448979591836737e-05,
+      "loss": 0.6119,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2653061224489794e-05,
+      "loss": 0.6986,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2857142857142858e-05,
+      "loss": 0.692,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.306122448979592e-05,
+      "loss": 0.6856,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.326530612244898e-05,
+      "loss": 0.6196,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.346938775510204e-05,
+      "loss": 0.6679,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3673469387755103e-05,
+      "loss": 0.6477,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3877551020408164e-05,
+      "loss": 0.6444,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4081632653061224e-05,
+      "loss": 0.6783,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4285714285714288e-05,
+      "loss": 0.6396,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.448979591836735e-05,
+      "loss": 0.6124,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.469387755102041e-05,
+      "loss": 0.6587,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4897959183673473e-05,
+      "loss": 0.5716,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5102040816326533e-05,
+      "loss": 0.5758,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5306122448979594e-05,
+      "loss": 0.6681,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.551020408163265e-05,
+      "loss": 0.6341,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5714285714285714e-05,
+      "loss": 0.6587,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5918367346938775e-05,
+      "loss": 0.6725,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6122448979591835e-05,
+      "loss": 0.6467,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.63265306122449e-05,
+      "loss": 0.6414,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.653061224489796e-05,
+      "loss": 0.6471,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.673469387755102e-05,
+      "loss": 0.6586,
+      "step": 131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6938775510204084e-05,
+      "loss": 0.668,
+      "step": 132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.7142857142857144e-05,
+      "loss": 0.6294,
+      "step": 133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.7346938775510205e-05,
+      "loss": 0.622,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7551020408163265e-05,
+      "loss": 0.688,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.775510204081633e-05,
+      "loss": 0.6511,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.795918367346939e-05,
+      "loss": 0.6755,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.816326530612245e-05,
+      "loss": 0.6391,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8367346938775514e-05,
+      "loss": 0.6583,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 0.7017,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.877551020408163e-05,
+      "loss": 0.6286,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8979591836734695e-05,
+      "loss": 0.5708,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9183673469387756e-05,
+      "loss": 0.6201,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9387755102040816e-05,
+      "loss": 0.6154,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9591836734693876e-05,
+      "loss": 0.6682,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.979591836734694e-05,
+      "loss": 0.6626,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3e-05,
+      "loss": 0.5985,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999999670121339e-05,
+      "loss": 0.6091,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999986804855002e-05,
+      "loss": 0.6255,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999970310929192e-05,
+      "loss": 0.5828,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999947219443217e-05,
+      "loss": 0.6094,
+      "step": 151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999991753040723e-05,
+      "loss": 0.6563,
+      "step": 152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999881243834292e-05,
+      "loss": 0.589,
+      "step": 153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999838359740362e-05,
+      "loss": 0.629,
+      "step": 154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999978887814431e-05,
+      "loss": 0.6275,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999973279906788e-05,
+      "loss": 0.5656,
+      "step": 156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999670122535755e-05,
+      "loss": 0.6065,
+      "step": 157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999600848575503e-05,
+      "loss": 0.599,
+      "step": 158
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9999524977217583e-05,
+      "loss": 0.6654,
+      "step": 159
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9999442508495367e-05,
+      "loss": 0.6201,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999935344244514e-05,
+      "loss": 0.6312,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9999257779106063e-05,
+      "loss": 0.6237,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999915551852022e-05,
+      "loss": 0.6475,
+      "step": 163
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9999046660732585e-05,
+      "loss": 0.6278,
+      "step": 164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998931205791042e-05,
+      "loss": 0.6338,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998809153746366e-05,
+      "loss": 0.5836,
+      "step": 166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999868050465225e-05,
+      "loss": 0.5909,
+      "step": 167
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998545258565275e-05,
+      "loss": 0.6456,
+      "step": 168
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998403415544917e-05,
+      "loss": 0.5795,
+      "step": 169
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998254975653577e-05,
+      "loss": 0.5642,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998099938956545e-05,
+      "loss": 0.6588,
+      "step": 171
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997938305522003e-05,
+      "loss": 0.6102,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997770075421044e-05,
+      "loss": 0.5767,
+      "step": 173
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999759524872767e-05,
+      "loss": 0.5887,
+      "step": 174
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997413825518772e-05,
+      "loss": 0.6104,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997225805874146e-05,
+      "loss": 0.5795,
+      "step": 176
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997031189876495e-05,
+      "loss": 0.6633,
+      "step": 177
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996829977611414e-05,
+      "loss": 0.6965,
+      "step": 178
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.99966221691674e-05,
+      "loss": 0.6035,
+      "step": 179
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996407764635865e-05,
+      "loss": 0.599,
+      "step": 180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996186764111103e-05,
+      "loss": 0.5676,
+      "step": 181
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9995959167690328e-05,
+      "loss": 0.6154,
+      "step": 182
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9995724975473636e-05,
+      "loss": 0.5714,
+      "step": 183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9995484187564042e-05,
+      "loss": 0.5895,
+      "step": 184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9995236804067447e-05,
+      "loss": 0.6074,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999498282509266e-05,
+      "loss": 0.5382,
+      "step": 186
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9994722250751395e-05,
+      "loss": 0.623,
+      "step": 187
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999445508115826e-05,
+      "loss": 0.6646,
+      "step": 188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999418131643077e-05,
+      "loss": 0.5711,
+      "step": 189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9993900956689335e-05,
+      "loss": 0.5865,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9993614002057262e-05,
+      "loss": 0.5858,
+      "step": 191
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999332045266077e-05,
+      "loss": 0.616,
+      "step": 192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999302030862898e-05,
+      "loss": 0.5956,
+      "step": 193
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9992713570093896e-05,
+      "loss": 0.6366,
+      "step": 194
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9992400237190433e-05,
+      "loss": 0.5715,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9992080310056413e-05,
+      "loss": 0.574,
+      "step": 196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9991753788832552e-05,
+      "loss": 0.5842,
+      "step": 197
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9991420673662464e-05,
+      "loss": 0.6359,
+      "step": 198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9991080964692664e-05,
+      "loss": 0.6448,
+      "step": 199
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9990734662072577e-05,
+      "loss": 0.553,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999038176595451e-05,
+      "loss": 0.6221,
+      "step": 201
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9990022276493685e-05,
+      "loss": 0.5587,
+      "step": 202
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998965619384822e-05,
+      "loss": 0.5592,
+      "step": 203
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998928351817913e-05,
+      "loss": 0.622,
+      "step": 204
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998890424965033e-05,
+      "loss": 0.6161,
+      "step": 205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9988518388428644e-05,
+      "loss": 0.5935,
+      "step": 206
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9988125934683788e-05,
+      "loss": 0.5258,
+      "step": 207
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998772688858837e-05,
+      "loss": 0.556,
+      "step": 208
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9987321250317908e-05,
+      "loss": 0.6054,
+      "step": 209
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9986909020050822e-05,
+      "loss": 0.5906,
+      "step": 210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9986490197968426e-05,
+      "loss": 0.5402,
+      "step": 211
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998606478425493e-05,
+      "loss": 0.6542,
+      "step": 212
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998563277909745e-05,
+      "loss": 0.5686,
+      "step": 213
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998519418268599e-05,
+      "loss": 0.6588,
+      "step": 214
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998474899521348e-05,
+      "loss": 0.591,
+      "step": 215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9984297216875714e-05,
+      "loss": 0.5735,
+      "step": 216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998383884787141e-05,
+      "loss": 0.6133,
+      "step": 217
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998337388840217e-05,
+      "loss": 0.5845,
+      "step": 218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9982902338672506e-05,
+      "loss": 0.505,
+      "step": 219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9982424198889823e-05,
+      "loss": 0.5717,
+      "step": 220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998193946926442e-05,
+      "loss": 0.5015,
+      "step": 221
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9981448150009506e-05,
+      "loss": 0.6031,
+      "step": 222
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998095024134118e-05,
+      "loss": 0.579,
+      "step": 223
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9980445743478438e-05,
+      "loss": 0.513,
+      "step": 224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9979934656643184e-05,
+      "loss": 0.5497,
+      "step": 225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9979416981060204e-05,
+      "loss": 0.608,
+      "step": 226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.99788927169572e-05,
+      "loss": 0.5416,
+      "step": 227
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.997836186456476e-05,
+      "loss": 0.5174,
+      "step": 228
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9977824424116378e-05,
+      "loss": 0.5119,
+      "step": 229
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.997728039584843e-05,
+      "loss": 0.5411,
+      "step": 230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.997672978000021e-05,
+      "loss": 0.5266,
+      "step": 231
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9976172576813895e-05,
+      "loss": 0.5268,
+      "step": 232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997560878653457e-05,
+      "loss": 0.5147,
+      "step": 233
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9975038409410205e-05,
+      "loss": 0.5997,
+      "step": 234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9974461445691673e-05,
+      "loss": 0.5644,
+      "step": 235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997387789563275e-05,
+      "loss": 0.576,
+      "step": 236
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9973287759490108e-05,
+      "loss": 0.6072,
+      "step": 237
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9972691037523303e-05,
+      "loss": 0.6015,
+      "step": 238
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997208772999479e-05,
+      "loss": 0.5163,
+      "step": 239
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9971477837169943e-05,
+      "loss": 0.487,
+      "step": 240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9970861359317002e-05,
+      "loss": 0.5919,
+      "step": 241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997023829670713e-05,
+      "loss": 0.6005,
+      "step": 242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9969608649614362e-05,
+      "loss": 0.5531,
+      "step": 243
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9968972418315648e-05,
+      "loss": 0.5703,
+      "step": 244
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9968329603090825e-05,
+      "loss": 0.548,
+      "step": 245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9967680204222628e-05,
+      "loss": 0.5396,
+      "step": 246
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9967024221996685e-05,
+      "loss": 0.5347,
+      "step": 247
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.996636165670153e-05,
+      "loss": 0.5464,
+      "step": 248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9965692508628572e-05,
+      "loss": 0.5886,
+      "step": 249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9965016778072135e-05,
+      "loss": 0.5803,
+      "step": 250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9964334465329435e-05,
+      "loss": 0.6203,
+      "step": 251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9963645570700573e-05,
+      "loss": 0.563,
+      "step": 252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9962950094488552e-05,
+      "loss": 0.5992,
+      "step": 253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.996224803699927e-05,
+      "loss": 0.5265,
+      "step": 254
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.996153939854152e-05,
+      "loss": 0.574,
+      "step": 255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9960824179426984e-05,
+      "loss": 0.5292,
+      "step": 256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.996010237997025e-05,
+      "loss": 0.562,
+      "step": 257
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9959374000488788e-05,
+      "loss": 0.5087,
+      "step": 258
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9958639041302968e-05,
+      "loss": 0.5652,
+      "step": 259
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.995789750273605e-05,
+      "loss": 0.5273,
+      "step": 260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9957149385114198e-05,
+      "loss": 0.572,
+      "step": 261
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9956394688766454e-05,
+      "loss": 0.4827,
+      "step": 262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.995563341402477e-05,
+      "loss": 0.5297,
+      "step": 263
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.995486556122398e-05,
+      "loss": 0.5817,
+      "step": 264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9954091130701816e-05,
+      "loss": 0.5066,
+      "step": 265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9953310122798907e-05,
+      "loss": 0.5413,
+      "step": 266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.995252253785876e-05,
+      "loss": 0.504,
+      "step": 267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9951728376227793e-05,
+      "loss": 0.5172,
+      "step": 268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.99509276382553e-05,
+      "loss": 0.4943,
+      "step": 269
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9950120324293484e-05,
+      "loss": 0.4926,
+      "step": 270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994930643469743e-05,
+      "loss": 0.4951,
+      "step": 271
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994848596982512e-05,
+      "loss": 0.5234,
+      "step": 272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9947658930037426e-05,
+      "loss": 0.576,
+      "step": 273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9946825315698105e-05,
+      "loss": 0.5613,
+      "step": 274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994598512717382e-05,
+      "loss": 0.5535,
+      "step": 275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994513836483412e-05,
+      "loss": 0.5759,
+      "step": 276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9944285029051427e-05,
+      "loss": 0.5618,
+      "step": 277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9943425120201087e-05,
+      "loss": 0.499,
+      "step": 278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9942558638661316e-05,
+      "loss": 0.5702,
+      "step": 279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9941685584813228e-05,
+      "loss": 0.56,
+      "step": 280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.994080595904082e-05,
+      "loss": 0.5115,
+      "step": 281
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9939919761730993e-05,
+      "loss": 0.5421,
+      "step": 282
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9939026993273518e-05,
+      "loss": 0.446,
+      "step": 283
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9938127654061082e-05,
+      "loss": 0.5721,
+      "step": 284
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.993722174448924e-05,
+      "loss": 0.5655,
+      "step": 285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.993630926495645e-05,
+      "loss": 0.5689,
+      "step": 286
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9935390215864057e-05,
+      "loss": 0.5374,
+      "step": 287
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9934464597616287e-05,
+      "loss": 0.5366,
+      "step": 288
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9933532410620266e-05,
+      "loss": 0.5253,
+      "step": 289
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9932593655286006e-05,
+      "loss": 0.5043,
+      "step": 290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9931648332026407e-05,
+      "loss": 0.5673,
+      "step": 291
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.993069644125726e-05,
+      "loss": 0.5238,
+      "step": 292
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.992973798339724e-05,
+      "loss": 0.5435,
+      "step": 293
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9928772958867917e-05,
+      "loss": 0.4834,
+      "step": 294
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9927801368093744e-05,
+      "loss": 0.4676,
+      "step": 295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9926823211502063e-05,
+      "loss": 0.48,
+      "step": 296
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.99258384895231e-05,
+      "loss": 0.5781,
+      "step": 297
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.992484720258998e-05,
+      "loss": 0.4781,
+      "step": 298
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.992384935113871e-05,
+      "loss": 0.5244,
+      "step": 299
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9922844935608177e-05,
+      "loss": 0.4568,
+      "step": 300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9921833956440167e-05,
+      "loss": 0.5294,
+      "step": 301
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9920816414079344e-05,
+      "loss": 0.5107,
+      "step": 302
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9919792308973264e-05,
+      "loss": 0.5379,
+      "step": 303
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9918761641572365e-05,
+      "loss": 0.516,
+      "step": 304
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9917724412329973e-05,
+      "loss": 0.4833,
+      "step": 305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9916680621702308e-05,
+      "loss": 0.4571,
+      "step": 306
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.991563027014846e-05,
+      "loss": 0.5044,
+      "step": 307
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9914573358130425e-05,
+      "loss": 0.4888,
+      "step": 308
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.991350988611306e-05,
+      "loss": 0.4873,
+      "step": 309
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.991243985456413e-05,
+      "loss": 0.4893,
+      "step": 310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9911363263954272e-05,
+      "loss": 0.5097,
+      "step": 311
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9910280114757012e-05,
+      "loss": 0.494,
+      "step": 312
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.990919040744876e-05,
+      "loss": 0.5175,
+      "step": 313
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9908094142508817e-05,
+      "loss": 0.5351,
+      "step": 314
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.990699132041935e-05,
+      "loss": 0.5202,
+      "step": 315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9905881941665436e-05,
+      "loss": 0.5503,
+      "step": 316
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9904766006735012e-05,
+      "loss": 0.5144,
+      "step": 317
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9903643516118913e-05,
+      "loss": 0.4456,
+      "step": 318
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.990251447031085e-05,
+      "loss": 0.4938,
+      "step": 319
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9901378869807428e-05,
+      "loss": 0.5737,
+      "step": 320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.990023671510812e-05,
+      "loss": 0.5698,
+      "step": 321
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9899088006715294e-05,
+      "loss": 0.5812,
+      "step": 322
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.989793274513419e-05,
+      "loss": 0.5164,
+      "step": 323
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9896770930872945e-05,
+      "loss": 0.4716,
+      "step": 324
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.989560256444256e-05,
+      "loss": 0.5525,
+      "step": 325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9894427646356934e-05,
+      "loss": 0.5509,
+      "step": 326
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9893246177132837e-05,
+      "loss": 0.485,
+      "step": 327
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.989205815728993e-05,
+      "loss": 0.5173,
+      "step": 328
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.989086358735074e-05,
+      "loss": 0.5225,
+      "step": 329
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.988966246784069e-05,
+      "loss": 0.5018,
+      "step": 330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.988845479928808e-05,
+      "loss": 0.4755,
+      "step": 331
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9887240582224085e-05,
+      "loss": 0.4202,
+      "step": 332
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9886019817182768e-05,
+      "loss": 0.425,
+      "step": 333
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9884792504701065e-05,
+      "loss": 0.4397,
+      "step": 334
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9883558645318798e-05,
+      "loss": 0.4731,
+      "step": 335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.988231823957866e-05,
+      "loss": 0.4168,
+      "step": 336
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9881071288026234e-05,
+      "loss": 0.5133,
+      "step": 337
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.987981779120997e-05,
+      "loss": 0.4844,
+      "step": 338
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9878557749681216e-05,
+      "loss": 0.4823,
+      "step": 339
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9877291163994176e-05,
+      "loss": 0.4742,
+      "step": 340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.987601803470595e-05,
+      "loss": 0.4967,
+      "step": 341
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.98747383623765e-05,
+      "loss": 0.5442,
+      "step": 342
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9873452147568684e-05,
+      "loss": 0.4632,
+      "step": 343
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.987215939084822e-05,
+      "loss": 0.485,
+      "step": 344
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9870860092783724e-05,
+      "loss": 0.4862,
+      "step": 345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.986955425394666e-05,
+      "loss": 0.4601,
+      "step": 346
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9868241874911402e-05,
+      "loss": 0.5248,
+      "step": 347
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9866922956255173e-05,
+      "loss": 0.4633,
+      "step": 348
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9865597498558086e-05,
+      "loss": 0.4835,
+      "step": 349
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9864265502403134e-05,
+      "loss": 0.5096,
+      "step": 350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9862926968376173e-05,
+      "loss": 0.4717,
+      "step": 351
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9861581897065947e-05,
+      "loss": 0.51,
+      "step": 352
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.986023028906407e-05,
+      "loss": 0.5459,
+      "step": 353
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9858872144965023e-05,
+      "loss": 0.5254,
+      "step": 354
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.985750746536618e-05,
+      "loss": 0.5274,
+      "step": 355
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.985613625086777e-05,
+      "loss": 0.4983,
+      "step": 356
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9854758502072907e-05,
+      "loss": 0.4342,
+      "step": 357
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9853374219587583e-05,
+      "loss": 0.429,
+      "step": 358
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9851983404020657e-05,
+      "loss": 0.4659,
+      "step": 359
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.985058605598386e-05,
+      "loss": 0.5355,
+      "step": 360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9849182176091797e-05,
+      "loss": 0.5017,
+      "step": 361
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9847771764961956e-05,
+      "loss": 0.4972,
+      "step": 362
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9846354823214683e-05,
+      "loss": 0.484,
+      "step": 363
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9844931351473202e-05,
+      "loss": 0.5091,
+      "step": 364
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9843501350363614e-05,
+      "loss": 0.4771,
+      "step": 365
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9842064820514895e-05,
+      "loss": 0.4815,
+      "step": 366
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.984062176255887e-05,
+      "loss": 0.4947,
+      "step": 367
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9839172177130263e-05,
+      "loss": 0.4625,
+      "step": 368
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9837716064866655e-05,
+      "loss": 0.4828,
+      "step": 369
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.983625342640849e-05,
+      "loss": 0.4959,
+      "step": 370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.983478426239911e-05,
+      "loss": 0.5007,
+      "step": 371
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9833308573484697e-05,
+      "loss": 0.5214,
+      "step": 372
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9831826360314316e-05,
+      "loss": 0.4544,
+      "step": 373
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9830337623539905e-05,
+      "loss": 0.4936,
+      "step": 374
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9828842363816266e-05,
+      "loss": 0.5324,
+      "step": 375
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9827340581801067e-05,
+      "loss": 0.5527,
+      "step": 376
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.982583227815486e-05,
+      "loss": 0.483,
+      "step": 377
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9824317453541042e-05,
+      "loss": 0.4954,
+      "step": 378
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9822796108625898e-05,
+      "loss": 0.4201,
+      "step": 379
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9821268244078572e-05,
+      "loss": 0.462,
+      "step": 380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.981973386057108e-05,
+      "loss": 0.4722,
+      "step": 381
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9818192958778297e-05,
+      "loss": 0.4457,
+      "step": 382
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9816645539377975e-05,
+      "loss": 0.4423,
+      "step": 383
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9815091603050725e-05,
+      "loss": 0.4885,
+      "step": 384
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9813531150480028e-05,
+      "loss": 0.3923,
+      "step": 385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9811964182352234e-05,
+      "loss": 0.4642,
+      "step": 386
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9810390699356552e-05,
+      "loss": 0.4292,
+      "step": 387
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.980881070218506e-05,
+      "loss": 0.5248,
+      "step": 388
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9807224191532707e-05,
+      "loss": 0.4991,
+      "step": 389
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9805631168097293e-05,
+      "loss": 0.4785,
+      "step": 390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.980403163257949e-05,
+      "loss": 0.4948,
+      "step": 391
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9802425585682845e-05,
+      "loss": 0.4769,
+      "step": 392
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9800813028113752e-05,
+      "loss": 0.4842,
+      "step": 393
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9799193960581473e-05,
+      "loss": 0.4418,
+      "step": 394
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.979756838379814e-05,
+      "loss": 0.4309,
+      "step": 395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9795936298478746e-05,
+      "loss": 0.4839,
+      "step": 396
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9794297705341136e-05,
+      "loss": 0.539,
+      "step": 397
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9792652605106032e-05,
+      "loss": 0.5266,
+      "step": 398
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.979100099849701e-05,
+      "loss": 0.4636,
+      "step": 399
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.978934288624051e-05,
+      "loss": 0.5306,
+      "step": 400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9787678269065836e-05,
+      "loss": 0.4728,
+      "step": 401
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9786007147705153e-05,
+      "loss": 0.4731,
+      "step": 402
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.978432952289347e-05,
+      "loss": 0.5173,
+      "step": 403
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9782645395368686e-05,
+      "loss": 0.4373,
+      "step": 404
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9780954765871537e-05,
+      "loss": 0.4988,
+      "step": 405
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9779257635145623e-05,
+      "loss": 0.4946,
+      "step": 406
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9777554003937414e-05,
+      "loss": 0.4257,
+      "step": 407
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.977584387299623e-05,
+      "loss": 0.5074,
+      "step": 408
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9774127243074248e-05,
+      "loss": 0.4526,
+      "step": 409
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.977240411492652e-05,
+      "loss": 0.4852,
+      "step": 410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9770674489310925e-05,
+      "loss": 0.5197,
+      "step": 411
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9768938366988232e-05,
+      "loss": 0.4719,
+      "step": 412
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9767195748722053e-05,
+      "loss": 0.4694,
+      "step": 413
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.976544663527885e-05,
+      "loss": 0.4167,
+      "step": 414
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.976369102742796e-05,
+      "loss": 0.4636,
+      "step": 415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9761928925941558e-05,
+      "loss": 0.4485,
+      "step": 416
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9760160331594687e-05,
+      "loss": 0.4415,
+      "step": 417
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9758385245165242e-05,
+      "loss": 0.4311,
+      "step": 418
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9756603667433977e-05,
+      "loss": 0.4975,
+      "step": 419
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9754815599184496e-05,
+      "loss": 0.4442,
+      "step": 420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.975302104120326e-05,
+      "loss": 0.4443,
+      "step": 421
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.975121999427958e-05,
+      "loss": 0.4988,
+      "step": 422
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9749412459205624e-05,
+      "loss": 0.4881,
+      "step": 423
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.974759843677643e-05,
+      "loss": 0.4285,
+      "step": 424
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9745777927789854e-05,
+      "loss": 0.3835,
+      "step": 425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.974395093304664e-05,
+      "loss": 0.4111,
+      "step": 426
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9742117453350364e-05,
+      "loss": 0.3784,
+      "step": 427
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9740277489507463e-05,
+      "loss": 0.4741,
+      "step": 428
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.973843104232722e-05,
+      "loss": 0.4709,
+      "step": 429
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9736578112621777e-05,
+      "loss": 0.4447,
+      "step": 430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9734718701206126e-05,
+      "loss": 0.497,
+      "step": 431
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9732852808898097e-05,
+      "loss": 0.4894,
+      "step": 432
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9730980436518384e-05,
+      "loss": 0.4386,
+      "step": 433
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9729101584890537e-05,
+      "loss": 0.4008,
+      "step": 434
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9727216254840934e-05,
+      "loss": 0.4297,
+      "step": 435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9725324447198824e-05,
+      "loss": 0.5257,
+      "step": 436
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9723426162796295e-05,
+      "loss": 0.4425,
+      "step": 437
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9721521402468285e-05,
+      "loss": 0.4663,
+      "step": 438
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9719610167052576e-05,
+      "loss": 0.3697,
+      "step": 439
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9717692457389806e-05,
+      "loss": 0.4447,
+      "step": 440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9715768274323458e-05,
+      "loss": 0.5072,
+      "step": 441
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.971383761869986e-05,
+      "loss": 0.4233,
+      "step": 442
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9711900491368186e-05,
+      "loss": 0.4821,
+      "step": 443
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.970995689318046e-05,
+      "loss": 0.4076,
+      "step": 444
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.970800682499155e-05,
+      "loss": 0.4545,
+      "step": 445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9706050287659175e-05,
+      "loss": 0.4013,
+      "step": 446
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9704087282043893e-05,
+      "loss": 0.399,
+      "step": 447
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.97021178090091e-05,
+      "loss": 0.4313,
+      "step": 448
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9700141869421057e-05,
+      "loss": 0.4513,
+      "step": 449
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.969815946414886e-05,
+      "loss": 0.4504,
+      "step": 450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9696170594064433e-05,
+      "loss": 0.4404,
+      "step": 451
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9694175260042567e-05,
+      "loss": 0.4253,
+      "step": 452
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9692173462960883e-05,
+      "loss": 0.4164,
+      "step": 453
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.969016520369985e-05,
+      "loss": 0.4057,
+      "step": 454
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.968815048314277e-05,
+      "loss": 0.5273,
+      "step": 455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9686129302175802e-05,
+      "loss": 0.4402,
+      "step": 456
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9684101661687937e-05,
+      "loss": 0.5798,
+      "step": 457
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.968206756257101e-05,
+      "loss": 0.4358,
+      "step": 458
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.968002700571969e-05,
+      "loss": 0.4634,
+      "step": 459
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9677979992031502e-05,
+      "loss": 0.3801,
+      "step": 460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9675926522406787e-05,
+      "loss": 0.468,
+      "step": 461
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.967386659774875e-05,
+      "loss": 0.47,
+      "step": 462
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9671800218963424e-05,
+      "loss": 0.3433,
+      "step": 463
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9669727386959676e-05,
+      "loss": 0.4963,
+      "step": 464
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9667648102649214e-05,
+      "loss": 0.4772,
+      "step": 465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9665562366946598e-05,
+      "loss": 0.501,
+      "step": 466
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.966347018076921e-05,
+      "loss": 0.4175,
+      "step": 467
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.966137154503727e-05,
+      "loss": 0.457,
+      "step": 468
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9659266460673837e-05,
+      "loss": 0.4479,
+      "step": 469
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.965715492860481e-05,
+      "loss": 0.4616,
+      "step": 470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.965503694975893e-05,
+      "loss": 0.4466,
+      "step": 471
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9652912525067746e-05,
+      "loss": 0.3744,
+      "step": 472
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.965078165546568e-05,
+      "loss": 0.4069,
+      "step": 473
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9648644341889965e-05,
+      "loss": 0.4076,
+      "step": 474
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.964650058528066e-05,
+      "loss": 0.4005,
+      "step": 475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.964435038658069e-05,
+      "loss": 0.3874,
+      "step": 476
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9642193746735782e-05,
+      "loss": 0.4207,
+      "step": 477
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.964003066669451e-05,
+      "loss": 0.4562,
+      "step": 478
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.963786114740829e-05,
+      "loss": 0.4483,
+      "step": 479
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9635685189831343e-05,
+      "loss": 0.412,
+      "step": 480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9633502794920752e-05,
+      "loss": 0.3941,
+      "step": 481
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9631313963636412e-05,
+      "loss": 0.4359,
+      "step": 482
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.962911869694105e-05,
+      "loss": 0.4285,
+      "step": 483
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9626916995800237e-05,
+      "loss": 0.4176,
+      "step": 484
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.962470886118236e-05,
+      "loss": 0.4185,
+      "step": 485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9622494294058645e-05,
+      "loss": 0.4405,
+      "step": 486
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9620273295403138e-05,
+      "loss": 0.4321,
+      "step": 487
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9618045866192724e-05,
+      "loss": 0.4169,
+      "step": 488
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.961581200740711e-05,
+      "loss": 0.4187,
+      "step": 489
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.961357172002883e-05,
+      "loss": 0.4028,
+      "step": 490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.961132500504325e-05,
+      "loss": 0.4594,
+      "step": 491
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9609071863438557e-05,
+      "loss": 0.4361,
+      "step": 492
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9606812296205777e-05,
+      "loss": 0.5088,
+      "step": 493
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9604546304338748e-05,
+      "loss": 0.4331,
+      "step": 494
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.960227388883414e-05,
+      "loss": 0.489,
+      "step": 495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9599995050691453e-05,
+      "loss": 0.3895,
+      "step": 496
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9597709790913004e-05,
+      "loss": 0.4084,
+      "step": 497
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9595418110503932e-05,
+      "loss": 0.431,
+      "step": 498
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9593120010472215e-05,
+      "loss": 0.4112,
+      "step": 499
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.959081549182864e-05,
+      "loss": 0.4414,
+      "step": 500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9588504555586824e-05,
+      "loss": 0.4618,
+      "step": 501
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9586187202763203e-05,
+      "loss": 0.4265,
+      "step": 502
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9583863434377044e-05,
+      "loss": 0.423,
+      "step": 503
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9581533251450416e-05,
+      "loss": 0.3768,
+      "step": 504
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.957919665500824e-05,
+      "loss": 0.3978,
+      "step": 505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9576853646078222e-05,
+      "loss": 0.3886,
+      "step": 506
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9574504225690923e-05,
+      "loss": 0.415,
+      "step": 507
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9572148394879695e-05,
+      "loss": 0.3865,
+      "step": 508
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.956978615468073e-05,
+      "loss": 0.4406,
+      "step": 509
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9567417506133032e-05,
+      "loss": 0.4702,
+      "step": 510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.956504245027842e-05,
+      "loss": 0.4296,
+      "step": 511
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.956266098816153e-05,
+      "loss": 0.3883,
+      "step": 512
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9560273120829828e-05,
+      "loss": 0.431,
+      "step": 513
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.955787884933359e-05,
+      "loss": 0.382,
+      "step": 514
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.95554781747259e-05,
+      "loss": 0.429,
+      "step": 515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9553071098062675e-05,
+      "loss": 0.4444,
+      "step": 516
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9550657620402632e-05,
+      "loss": 0.3754,
+      "step": 517
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9548237742807316e-05,
+      "loss": 0.4344,
+      "step": 518
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9545811466341078e-05,
+      "loss": 0.3799,
+      "step": 519
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.954337879207109e-05,
+      "loss": 0.4145,
+      "step": 520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9540939721067334e-05,
+      "loss": 0.4978,
+      "step": 521
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9538494254402604e-05,
+      "loss": 0.4681,
+      "step": 522
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9536042393152512e-05,
+      "loss": 0.398,
+      "step": 523
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9533584138395488e-05,
+      "loss": 0.4582,
+      "step": 524
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.953111949121275e-05,
+      "loss": 0.4135,
+      "step": 525
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9528648452688358e-05,
+      "loss": 0.4455,
+      "step": 526
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.952617102390916e-05,
+      "loss": 0.4613,
+      "step": 527
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.952368720596483e-05,
+      "loss": 0.3665,
+      "step": 528
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9521196999947846e-05,
+      "loss": 0.421,
+      "step": 529
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.951870040695349e-05,
+      "loss": 0.4107,
+      "step": 530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9516197428079866e-05,
+      "loss": 0.3627,
+      "step": 531
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9513688064427876e-05,
+      "loss": 0.4384,
+      "step": 532
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.951117231710123e-05,
+      "loss": 0.4019,
+      "step": 533
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9508650187206456e-05,
+      "loss": 0.4333,
+      "step": 534
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.950612167585288e-05,
+      "loss": 0.4206,
+      "step": 535
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9503586784152645e-05,
+      "loss": 0.4618,
+      "step": 536
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.950104551322069e-05,
+      "loss": 0.3909,
+      "step": 537
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.949849786417475e-05,
+      "loss": 0.386,
+      "step": 538
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.94959438381354e-05,
+      "loss": 0.387,
+      "step": 539
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.949338343622598e-05,
+      "loss": 0.3752,
+      "step": 540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.949081665957266e-05,
+      "loss": 0.4519,
+      "step": 541
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.948824350930441e-05,
+      "loss": 0.4593,
+      "step": 542
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9485663986552992e-05,
+      "loss": 0.4295,
+      "step": 543
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9483078092452984e-05,
+      "loss": 0.4063,
+      "step": 544
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.948048582814176e-05,
+      "loss": 0.4095,
+      "step": 545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.94778871947595e-05,
+      "loss": 0.4314,
+      "step": 546
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9475282193449175e-05,
+      "loss": 0.3793,
+      "step": 547
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9472670825356572e-05,
+      "loss": 0.4159,
+      "step": 548
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9470053091630265e-05,
+      "loss": 0.4084,
+      "step": 549
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9467428993421635e-05,
+      "loss": 0.4643,
+      "step": 550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9464798531884863e-05,
+      "loss": 0.3886,
+      "step": 551
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9462161708176922e-05,
+      "loss": 0.3752,
+      "step": 552
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9459518523457587e-05,
+      "loss": 0.4426,
+      "step": 553
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.945686897888944e-05,
+      "loss": 0.4304,
+      "step": 554
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9454213075637842e-05,
+      "loss": 0.3899,
+      "step": 555
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9451550814870968e-05,
+      "loss": 0.3977,
+      "step": 556
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.944888219775978e-05,
+      "loss": 0.3342,
+      "step": 557
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9446207225478035e-05,
+      "loss": 0.4355,
+      "step": 558
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.944352589920229e-05,
+      "loss": 0.4372,
+      "step": 559
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9440838220111898e-05,
+      "loss": 0.3883,
+      "step": 560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9438144189388997e-05,
+      "loss": 0.4106,
+      "step": 561
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.943544380821853e-05,
+      "loss": 0.4733,
+      "step": 562
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9432737077788228e-05,
+      "loss": 0.3774,
+      "step": 563
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9430023999288608e-05,
+      "loss": 0.4694,
+      "step": 564
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9427304573912987e-05,
+      "loss": 0.3717,
+      "step": 565
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.942457880285748e-05,
+      "loss": 0.3503,
+      "step": 566
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9421846687320978e-05,
+      "loss": 0.4356,
+      "step": 567
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.941910822850517e-05,
+      "loss": 0.4298,
+      "step": 568
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.941636342761454e-05,
+      "loss": 0.3913,
+      "step": 569
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.941361228585635e-05,
+      "loss": 0.4187,
+      "step": 570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9410854804440662e-05,
+      "loss": 0.4747,
+      "step": 571
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9408090984580323e-05,
+      "loss": 0.407,
+      "step": 572
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.940532082749096e-05,
+      "loss": 0.4039,
+      "step": 573
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9402544334390995e-05,
+      "loss": 0.4394,
+      "step": 574
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9399761506501643e-05,
+      "loss": 0.4306,
+      "step": 575
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.939697234504689e-05,
+      "loss": 0.3771,
+      "step": 576
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.939417685125352e-05,
+      "loss": 0.4129,
+      "step": 577
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.93913750263511e-05,
+      "loss": 0.3994,
+      "step": 578
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.938856687157197e-05,
+      "loss": 0.4298,
+      "step": 579
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9385752388151277e-05,
+      "loss": 0.3958,
+      "step": 580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9382931577326934e-05,
+      "loss": 0.3775,
+      "step": 581
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9380104440339633e-05,
+      "loss": 0.3971,
+      "step": 582
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.937727097843287e-05,
+      "loss": 0.4622,
+      "step": 583
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9374431192852898e-05,
+      "loss": 0.4776,
+      "step": 584
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9371585084848765e-05,
+      "loss": 0.3151,
+      "step": 585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9368732655672307e-05,
+      "loss": 0.4138,
+      "step": 586
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9365873906578126e-05,
+      "loss": 0.3508,
+      "step": 587
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9363008838823607e-05,
+      "loss": 0.3732,
+      "step": 588
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9360137453668917e-05,
+      "loss": 0.359,
+      "step": 589
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9357259752377007e-05,
+      "loss": 0.4267,
+      "step": 590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9354375736213587e-05,
+      "loss": 0.4222,
+      "step": 591
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.935148540644717e-05,
+      "loss": 0.3714,
+      "step": 592
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.934858876434903e-05,
+      "loss": 0.3791,
+      "step": 593
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9345685811193223e-05,
+      "loss": 0.3886,
+      "step": 594
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9342776548256567e-05,
+      "loss": 0.3513,
+      "step": 595
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.933986097681868e-05,
+      "loss": 0.382,
+      "step": 596
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.933693909816194e-05,
+      "loss": 0.3564,
+      "step": 597
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9334010913571493e-05,
+      "loss": 0.4541,
+      "step": 598
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.933107642433528e-05,
+      "loss": 0.4269,
+      "step": 599
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9328135631743983e-05,
+      "loss": 0.4224,
+      "step": 600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9325188537091095e-05,
+      "loss": 0.3737,
+      "step": 601
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9322235141672845e-05,
+      "loss": 0.3617,
+      "step": 602
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9319275446788257e-05,
+      "loss": 0.3975,
+      "step": 603
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9316309453739118e-05,
+      "loss": 0.3445,
+      "step": 604
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.931333716382998e-05,
+      "loss": 0.3807,
+      "step": 605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9310358578368176e-05,
+      "loss": 0.3727,
+      "step": 606
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9307373698663793e-05,
+      "loss": 0.361,
+      "step": 607
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9304382526029708e-05,
+      "loss": 0.4044,
+      "step": 608
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9301385061781542e-05,
+      "loss": 0.3861,
+      "step": 609
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9298381307237697e-05,
+      "loss": 0.409,
+      "step": 610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9295371263719342e-05,
+      "loss": 0.3975,
+      "step": 611
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.929235493255041e-05,
+      "loss": 0.3764,
+      "step": 612
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.928933231505759e-05,
+      "loss": 0.3893,
+      "step": 613
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9286303412570358e-05,
+      "loss": 0.4163,
+      "step": 614
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.928326822642093e-05,
+      "loss": 0.3802,
+      "step": 615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.92802267579443e-05,
+      "loss": 0.3492,
+      "step": 616
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9277179008478225e-05,
+      "loss": 0.4078,
+      "step": 617
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9274124979363218e-05,
+      "loss": 0.3919,
+      "step": 618
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9271064671942563e-05,
+      "loss": 0.3709,
+      "step": 619
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.926799808756229e-05,
+      "loss": 0.401,
+      "step": 620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.926492522757121e-05,
+      "loss": 0.3738,
+      "step": 621
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.926184609332088e-05,
+      "loss": 0.3638,
+      "step": 622
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.925876068616562e-05,
+      "loss": 0.3685,
+      "step": 623
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9255669007462513e-05,
+      "loss": 0.3996,
+      "step": 624
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9252571058571396e-05,
+      "loss": 0.409,
+      "step": 625
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9249466840854867e-05,
+      "loss": 0.3037,
+      "step": 626
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9246356355678277e-05,
+      "loss": 0.3715,
+      "step": 627
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9243239604409733e-05,
+      "loss": 0.4247,
+      "step": 628
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.924011658842011e-05,
+      "loss": 0.3631,
+      "step": 629
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9236987309083023e-05,
+      "loss": 0.3783,
+      "step": 630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9233851767774852e-05,
+      "loss": 0.3886,
+      "step": 631
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.923070996587473e-05,
+      "loss": 0.3502,
+      "step": 632
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.922756190476453e-05,
+      "loss": 0.4163,
+      "step": 633
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9224407585828905e-05,
+      "loss": 0.4183,
+      "step": 634
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9221247010455232e-05,
+      "loss": 0.4086,
+      "step": 635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9218080180033664e-05,
+      "loss": 0.3865,
+      "step": 636
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9214907095957086e-05,
+      "loss": 0.3739,
+      "step": 637
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9211727759621143e-05,
+      "loss": 0.2829,
+      "step": 638
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9208542172424234e-05,
+      "loss": 0.367,
+      "step": 639
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9205350335767496e-05,
+      "loss": 0.3851,
+      "step": 640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.920215225105482e-05,
+      "loss": 0.4418,
+      "step": 641
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.919894791969285e-05,
+      "loss": 0.3896,
+      "step": 642
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9195737343090973e-05,
+      "loss": 0.3898,
+      "step": 643
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9192520522661324e-05,
+      "loss": 0.3414,
+      "step": 644
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.918929745981878e-05,
+      "loss": 0.354,
+      "step": 645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.918606815598097e-05,
+      "loss": 0.3968,
+      "step": 646
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9182832612568263e-05,
+      "loss": 0.3612,
+      "step": 647
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9179590831003774e-05,
+      "loss": 0.3114,
+      "step": 648
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9176342812713366e-05,
+      "loss": 0.3703,
+      "step": 649
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9173088559125643e-05,
+      "loss": 0.4236,
+      "step": 650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9169828071671943e-05,
+      "loss": 0.4007,
+      "step": 651
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.916656135178636e-05,
+      "loss": 0.3808,
+      "step": 652
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9163288400905718e-05,
+      "loss": 0.3402,
+      "step": 653
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.916000922046958e-05,
+      "loss": 0.3759,
+      "step": 654
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.915672381192027e-05,
+      "loss": 0.3232,
+      "step": 655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9153432176702825e-05,
+      "loss": 0.4195,
+      "step": 656
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9150134316265037e-05,
+      "loss": 0.3975,
+      "step": 657
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.914683023205743e-05,
+      "loss": 0.3963,
+      "step": 658
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9143519925533262e-05,
+      "loss": 0.4025,
+      "step": 659
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.914020339814854e-05,
+      "loss": 0.3604,
+      "step": 660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.913688065136199e-05,
+      "loss": 0.3686,
+      "step": 661
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9133551686635096e-05,
+      "loss": 0.3657,
+      "step": 662
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9130216505432052e-05,
+      "loss": 0.3846,
+      "step": 663
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9126875109219798e-05,
+      "loss": 0.3706,
+      "step": 664
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.912352749946802e-05,
+      "loss": 0.362,
+      "step": 665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9120173677649115e-05,
+      "loss": 0.3813,
+      "step": 666
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.911681364523822e-05,
+      "loss": 0.3883,
+      "step": 667
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9113447403713216e-05,
+      "loss": 0.3805,
+      "step": 668
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9110074954554696e-05,
+      "loss": 0.3188,
+      "step": 669
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9106696299245986e-05,
+      "loss": 0.3488,
+      "step": 670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9103311439273167e-05,
+      "loss": 0.3685,
+      "step": 671
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9099920376125015e-05,
+      "loss": 0.3307,
+      "step": 672
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.909652311129305e-05,
+      "loss": 0.3494,
+      "step": 673
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9093119646271525e-05,
+      "loss": 0.392,
+      "step": 674
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9089709982557414e-05,
+      "loss": 0.3957,
+      "step": 675
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.908629412165041e-05,
+      "loss": 0.3845,
+      "step": 676
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9082872065052943e-05,
+      "loss": 0.352,
+      "step": 677
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9079443814270163e-05,
+      "loss": 0.4199,
+      "step": 678
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9076009370809952e-05,
+      "loss": 0.4158,
+      "step": 679
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9072568736182903e-05,
+      "loss": 0.3408,
+      "step": 680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.906912191190234e-05,
+      "loss": 0.3847,
+      "step": 681
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9065668899484308e-05,
+      "loss": 0.3671,
+      "step": 682
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9062209700447576e-05,
+      "loss": 0.381,
+      "step": 683
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.905874431631363e-05,
+      "loss": 0.3615,
+      "step": 684
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9055272748606678e-05,
+      "loss": 0.3565,
+      "step": 685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9051794998853648e-05,
+      "loss": 0.3405,
+      "step": 686
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.904831106858419e-05,
+      "loss": 0.3713,
+      "step": 687
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9044820959330663e-05,
+      "loss": 0.3016,
+      "step": 688
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.904132467262816e-05,
+      "loss": 0.3757,
+      "step": 689
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9037822210014472e-05,
+      "loss": 0.3666,
+      "step": 690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9034313573030125e-05,
+      "loss": 0.3707,
+      "step": 691
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9030798763218347e-05,
+      "loss": 0.33,
+      "step": 692
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9027277782125083e-05,
+      "loss": 0.3876,
+      "step": 693
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9023750631299e-05,
+      "loss": 0.377,
+      "step": 694
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.902021731229147e-05,
+      "loss": 0.3188,
+      "step": 695
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9016677826656584e-05,
+      "loss": 0.3839,
+      "step": 696
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.901313217595114e-05,
+      "loss": 0.2933,
+      "step": 697
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9009580361734656e-05,
+      "loss": 0.3644,
+      "step": 698
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9006022385569354e-05,
+      "loss": 0.3757,
+      "step": 699
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9002458249020162e-05,
+      "loss": 0.3558,
+      "step": 700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.899888795365473e-05,
+      "loss": 0.3548,
+      "step": 701
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8995311501043407e-05,
+      "loss": 0.2999,
+      "step": 702
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8991728892759257e-05,
+      "loss": 0.3567,
+      "step": 703
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8988140130378046e-05,
+      "loss": 0.4084,
+      "step": 704
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8984545215478246e-05,
+      "loss": 0.4027,
+      "step": 705
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.898094414964104e-05,
+      "loss": 0.3398,
+      "step": 706
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.897733693445032e-05,
+      "loss": 0.3134,
+      "step": 707
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.897372357149267e-05,
+      "loss": 0.4116,
+      "step": 708
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.897010406235739e-05,
+      "loss": 0.3879,
+      "step": 709
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.896647840863647e-05,
+      "loss": 0.4129,
+      "step": 710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.896284661192462e-05,
+      "loss": 0.3523,
+      "step": 711
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.895920867381924e-05,
+      "loss": 0.3526,
+      "step": 712
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.895556459592043e-05,
+      "loss": 0.3323,
+      "step": 713
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8951914379831e-05,
+      "loss": 0.3339,
+      "step": 714
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.894825802715646e-05,
+      "loss": 0.3316,
+      "step": 715
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8944595539505002e-05,
+      "loss": 0.3196,
+      "step": 716
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.894092691848753e-05,
+      "loss": 0.3558,
+      "step": 717
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.893725216571765e-05,
+      "loss": 0.3696,
+      "step": 718
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8933571282811654e-05,
+      "loss": 0.2969,
+      "step": 719
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8929884271388535e-05,
+      "loss": 0.4137,
+      "step": 720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8926191133069984e-05,
+      "loss": 0.3132,
+      "step": 721
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8922491869480386e-05,
+      "loss": 0.4014,
+      "step": 722
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8918786482246807e-05,
+      "loss": 0.3507,
+      "step": 723
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8915074972999034e-05,
+      "loss": 0.3451,
+      "step": 724
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8911357343369523e-05,
+      "loss": 0.3463,
+      "step": 725
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8907633594993426e-05,
+      "loss": 0.3497,
+      "step": 726
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8903903729508597e-05,
+      "loss": 0.3646,
+      "step": 727
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8900167748555567e-05,
+      "loss": 0.3496,
+      "step": 728
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8896425653777568e-05,
+      "loss": 0.4009,
+      "step": 729
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8892677446820518e-05,
+      "loss": 0.3575,
+      "step": 730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.888892312933301e-05,
+      "loss": 0.3778,
+      "step": 731
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.888516270296635e-05,
+      "loss": 0.3448,
+      "step": 732
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.888139616937451e-05,
+      "loss": 0.3398,
+      "step": 733
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8877623530214157e-05,
+      "loss": 0.3793,
+      "step": 734
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.887384478714464e-05,
+      "loss": 0.4225,
+      "step": 735
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8870059941828e-05,
+      "loss": 0.2807,
+      "step": 736
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.886626899592895e-05,
+      "loss": 0.3972,
+      "step": 737
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8862471951114896e-05,
+      "loss": 0.3886,
+      "step": 738
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8858668809055925e-05,
+      "loss": 0.3315,
+      "step": 739
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8854859571424803e-05,
+      "loss": 0.3086,
+      "step": 740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8851044239896972e-05,
+      "loss": 0.2998,
+      "step": 741
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8847222816150574e-05,
+      "loss": 0.3965,
+      "step": 742
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.884339530186641e-05,
+      "loss": 0.2735,
+      "step": 743
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.883956169872796e-05,
+      "loss": 0.4025,
+      "step": 744
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.88357220084214e-05,
+      "loss": 0.4224,
+      "step": 745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8831876232635568e-05,
+      "loss": 0.3442,
+      "step": 746
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8828024373061985e-05,
+      "loss": 0.3722,
+      "step": 747
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8824166431394844e-05,
+      "loss": 0.3478,
+      "step": 748
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8820302409331014e-05,
+      "loss": 0.3271,
+      "step": 749
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.881643230857004e-05,
+      "loss": 0.3636,
+      "step": 750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8812556130814144e-05,
+      "loss": 0.4074,
+      "step": 751
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8808673877768216e-05,
+      "loss": 0.2847,
+      "step": 752
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8804785551139817e-05,
+      "loss": 0.272,
+      "step": 753
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8800891152639182e-05,
+      "loss": 0.3505,
+      "step": 754
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8796990683979215e-05,
+      "loss": 0.2989,
+      "step": 755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8793084146875496e-05,
+      "loss": 0.3163,
+      "step": 756
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8789171543046263e-05,
+      "loss": 0.314,
+      "step": 757
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8785252874212434e-05,
+      "loss": 0.3523,
+      "step": 758
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8781328142097588e-05,
+      "loss": 0.3393,
+      "step": 759
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.877739734842797e-05,
+      "loss": 0.3042,
+      "step": 760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8773460494932497e-05,
+      "loss": 0.3351,
+      "step": 761
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.876951758334274e-05,
+      "loss": 0.3094,
+      "step": 762
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8765568615392954e-05,
+      "loss": 0.4233,
+      "step": 763
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8761613592820038e-05,
+      "loss": 0.3475,
+      "step": 764
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8757652517363563e-05,
+      "loss": 0.3602,
+      "step": 765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.875368539076576e-05,
+      "loss": 0.3364,
+      "step": 766
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.874971221477153e-05,
+      "loss": 0.3098,
+      "step": 767
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8745732991128416e-05,
+      "loss": 0.3359,
+      "step": 768
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8741747721586644e-05,
+      "loss": 0.3313,
+      "step": 769
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.873775640789908e-05,
+      "loss": 0.3836,
+      "step": 770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8733759051821263e-05,
+      "loss": 0.3246,
+      "step": 771
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8729755655111376e-05,
+      "loss": 0.2738,
+      "step": 772
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.872574621953027e-05,
+      "loss": 0.3301,
+      "step": 773
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8721730746841447e-05,
+      "loss": 0.3493,
+      "step": 774
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8717709238811062e-05,
+      "loss": 0.2927,
+      "step": 775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8713681697207934e-05,
+      "loss": 0.3393,
+      "step": 776
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.870964812380353e-05,
+      "loss": 0.2369,
+      "step": 777
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.870560852037196e-05,
+      "loss": 0.3679,
+      "step": 778
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.870156288869001e-05,
+      "loss": 0.3621,
+      "step": 779
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8697511230537094e-05,
+      "loss": 0.3448,
+      "step": 780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8693453547695285e-05,
+      "loss": 0.3365,
+      "step": 781
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8689389841949313e-05,
+      "loss": 0.3444,
+      "step": 782
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8685320115086552e-05,
+      "loss": 0.3991,
+      "step": 783
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8681244368897014e-05,
+      "loss": 0.3874,
+      "step": 784
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8677162605173372e-05,
+      "loss": 0.3097,
+      "step": 785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8673074825710947e-05,
+      "loss": 0.3143,
+      "step": 786
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8668981032307698e-05,
+      "loss": 0.3119,
+      "step": 787
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8664881226764227e-05,
+      "loss": 0.3313,
+      "step": 788
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8660775410883794e-05,
+      "loss": 0.3062,
+      "step": 789
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8656663586472286e-05,
+      "loss": 0.3773,
+      "step": 790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8652545755338243e-05,
+      "loss": 0.3901,
+      "step": 791
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8648421919292843e-05,
+      "loss": 0.2956,
+      "step": 792
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.864429208014991e-05,
+      "loss": 0.3631,
+      "step": 793
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.86401562397259e-05,
+      "loss": 0.2588,
+      "step": 794
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8636014399839924e-05,
+      "loss": 0.3647,
+      "step": 795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.863186656231371e-05,
+      "loss": 0.3501,
+      "step": 796
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.862771272897164e-05,
+      "loss": 0.3574,
+      "step": 797
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.862355290164073e-05,
+      "loss": 0.3546,
+      "step": 798
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8619387082150626e-05,
+      "loss": 0.3717,
+      "step": 799
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8615215272333623e-05,
+      "loss": 0.3853,
+      "step": 800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8611037474024637e-05,
+      "loss": 0.3809,
+      "step": 801
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.860685368906122e-05,
+      "loss": 0.3802,
+      "step": 802
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.860266391928357e-05,
+      "loss": 0.3108,
+      "step": 803
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8598468166534494e-05,
+      "loss": 0.3573,
+      "step": 804
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.859426643265946e-05,
+      "loss": 0.2839,
+      "step": 805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.859005871950654e-05,
+      "loss": 0.3011,
+      "step": 806
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8585845028926456e-05,
+      "loss": 0.2802,
+      "step": 807
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.858162536277254e-05,
+      "loss": 0.3436,
+      "step": 808
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.857739972290077e-05,
+      "loss": 0.3155,
+      "step": 809
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.857316811116974e-05,
+      "loss": 0.2588,
+      "step": 810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8568930529440677e-05,
+      "loss": 0.2998,
+      "step": 811
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.856468697957743e-05,
+      "loss": 0.3121,
+      "step": 812
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8560437463446474e-05,
+      "loss": 0.3305,
+      "step": 813
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8556181982916908e-05,
+      "loss": 0.2765,
+      "step": 814
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.855192053986046e-05,
+      "loss": 0.3606,
+      "step": 815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.854765313615147e-05,
+      "loss": 0.318,
+      "step": 816
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.85433797736669e-05,
+      "loss": 0.3026,
+      "step": 817
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8539100454286357e-05,
+      "loss": 0.3368,
+      "step": 818
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.853481517989203e-05,
+      "loss": 0.3269,
+      "step": 819
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8530523952368755e-05,
+      "loss": 0.321,
+      "step": 820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.852622677360398e-05,
+      "loss": 0.3538,
+      "step": 821
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.852192364548776e-05,
+      "loss": 0.2693,
+      "step": 822
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8517614569912784e-05,
+      "loss": 0.2959,
+      "step": 823
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8513299548774345e-05,
+      "loss": 0.3509,
+      "step": 824
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.850897858397035e-05,
+      "loss": 0.3105,
+      "step": 825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.850465167740133e-05,
+      "loss": 0.394,
+      "step": 826
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8500318830970422e-05,
+      "loss": 0.3477,
+      "step": 827
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8495980046583375e-05,
+      "loss": 0.361,
+      "step": 828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8491635326148558e-05,
+      "loss": 0.2788,
+      "step": 829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8487284671576937e-05,
+      "loss": 0.3118,
+      "step": 830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8482928084782104e-05,
+      "loss": 0.2914,
+      "step": 831
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.847856556768025e-05,
+      "loss": 0.3309,
+      "step": 832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8474197122190172e-05,
+      "loss": 0.2993,
+      "step": 833
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8469822750233287e-05,
+      "loss": 0.4292,
+      "step": 834
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.84654424537336e-05,
+      "loss": 0.363,
+      "step": 835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8461056234617746e-05,
+      "loss": 0.2818,
+      "step": 836
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8456664094814947e-05,
+      "loss": 0.2867,
+      "step": 837
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8452266036257027e-05,
+      "loss": 0.3415,
+      "step": 838
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.844786206087843e-05,
+      "loss": 0.2479,
+      "step": 839
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8443452170616188e-05,
+      "loss": 0.3105,
+      "step": 840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8439036367409938e-05,
+      "loss": 0.2921,
+      "step": 841
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.843461465320192e-05,
+      "loss": 0.2936,
+      "step": 842
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.843018702993698e-05,
+      "loss": 0.2755,
+      "step": 843
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8425753499562542e-05,
+      "loss": 0.3181,
+      "step": 844
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.842131406402865e-05,
+      "loss": 0.3294,
+      "step": 845
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.841686872528794e-05,
+      "loss": 0.2612,
+      "step": 846
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8412417485295637e-05,
+      "loss": 0.3273,
+      "step": 847
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.840796034600957e-05,
+      "loss": 0.3109,
+      "step": 848
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.840349730939016e-05,
+      "loss": 0.3161,
+      "step": 849
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8399028377400412e-05,
+      "loss": 0.326,
+      "step": 850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8394553552005943e-05,
+      "loss": 0.3288,
+      "step": 851
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8390072835174948e-05,
+      "loss": 0.3328,
+      "step": 852
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.838558622887822e-05,
+      "loss": 0.3575,
+      "step": 853
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8381093735089137e-05,
+      "loss": 0.2934,
+      "step": 854
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8376595355783665e-05,
+      "loss": 0.3014,
+      "step": 855
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8372091092940376e-05,
+      "loss": 0.3097,
+      "step": 856
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8367580948540408e-05,
+      "loss": 0.3212,
+      "step": 857
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8363064924567498e-05,
+      "loss": 0.3377,
+      "step": 858
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.835854302300796e-05,
+      "loss": 0.3385,
+      "step": 859
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.835401524585071e-05,
+      "loss": 0.3446,
+      "step": 860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8349481595087227e-05,
+      "loss": 0.3549,
+      "step": 861
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8344942072711587e-05,
+      "loss": 0.3153,
+      "step": 862
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8340396680720447e-05,
+      "loss": 0.3178,
+      "step": 863
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.833584542111305e-05,
+      "loss": 0.3395,
+      "step": 864
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.83312882958912e-05,
+      "loss": 0.3373,
+      "step": 865
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.83267253070593e-05,
+      "loss": 0.3284,
+      "step": 866
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8322156456624333e-05,
+      "loss": 0.368,
+      "step": 867
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.831758174659585e-05,
+      "loss": 0.374,
+      "step": 868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.831300117898598e-05,
+      "loss": 0.2408,
+      "step": 869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8308414755809437e-05,
+      "loss": 0.369,
+      "step": 870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8303822479083507e-05,
+      "loss": 0.3473,
+      "step": 871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.829922435082804e-05,
+      "loss": 0.357,
+      "step": 872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8294620373065473e-05,
+      "loss": 0.3361,
+      "step": 873
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8290010547820814e-05,
+      "loss": 0.3398,
+      "step": 874
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.828539487712164e-05,
+      "loss": 0.3081,
+      "step": 875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.82807733629981e-05,
+      "loss": 0.3297,
+      "step": 876
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.82761460074829e-05,
+      "loss": 0.3194,
+      "step": 877
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8271512812611346e-05,
+      "loss": 0.2801,
+      "step": 878
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.826687378042128e-05,
+      "loss": 0.3637,
+      "step": 879
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.826222891295314e-05,
+      "loss": 0.3374,
+      "step": 880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.82575782122499e-05,
+      "loss": 0.2857,
+      "step": 881
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.825292168035713e-05,
+      "loss": 0.3972,
+      "step": 882
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.824825931932294e-05,
+      "loss": 0.3556,
+      "step": 883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8243591131198018e-05,
+      "loss": 0.2642,
+      "step": 884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8238917118035618e-05,
+      "loss": 0.3801,
+      "step": 885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8234237281891538e-05,
+      "loss": 0.3229,
+      "step": 886
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8229551624824154e-05,
+      "loss": 0.3354,
+      "step": 887
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8224860148894405e-05,
+      "loss": 0.3345,
+      "step": 888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.822016285616577e-05,
+      "loss": 0.2906,
+      "step": 889
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8215459748704303e-05,
+      "loss": 0.2859,
+      "step": 890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8210750828578607e-05,
+      "loss": 0.3567,
+      "step": 891
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8206036097859853e-05,
+      "loss": 0.2624,
+      "step": 892
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.820131555862175e-05,
+      "loss": 0.3178,
+      "step": 893
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8196589212940577e-05,
+      "loss": 0.2603,
+      "step": 894
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.819185706289516e-05,
+      "loss": 0.2987,
+      "step": 895
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8187119110566882e-05,
+      "loss": 0.3533,
+      "step": 896
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8182375358039673e-05,
+      "loss": 0.3171,
+      "step": 897
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8177625807400012e-05,
+      "loss": 0.2791,
+      "step": 898
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.817287046073694e-05,
+      "loss": 0.2791,
+      "step": 899
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8168109320142042e-05,
+      "loss": 0.3453,
+      "step": 900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.816334238770944e-05,
+      "loss": 0.3535,
+      "step": 901
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8158569665535823e-05,
+      "loss": 0.2995,
+      "step": 902
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.815379115572041e-05,
+      "loss": 0.2682,
+      "step": 903
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8149006860364976e-05,
+      "loss": 0.3264,
+      "step": 904
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8144216781573832e-05,
+      "loss": 0.2742,
+      "step": 905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8139420921453843e-05,
+      "loss": 0.3479,
+      "step": 906
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.813461928211441e-05,
+      "loss": 0.3192,
+      "step": 907
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8129811865667482e-05,
+      "loss": 0.3114,
+      "step": 908
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.812499867422753e-05,
+      "loss": 0.3189,
+      "step": 909
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8120179709911593e-05,
+      "loss": 0.3186,
+      "step": 910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.811535497483923e-05,
+      "loss": 0.3942,
+      "step": 911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8110524471132553e-05,
+      "loss": 0.3361,
+      "step": 912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8105688200916186e-05,
+      "loss": 0.302,
+      "step": 913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.810084616631732e-05,
+      "loss": 0.3239,
+      "step": 914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8095998369465656e-05,
+      "loss": 0.2627,
+      "step": 915
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.809114481249345e-05,
+      "loss": 0.3261,
+      "step": 916
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.808628549753548e-05,
+      "loss": 0.2864,
+      "step": 917
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.808142042672905e-05,
+      "loss": 0.2916,
+      "step": 918
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.807654960221401e-05,
+      "loss": 0.3011,
+      "step": 919
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8071673026132732e-05,
+      "loss": 0.3082,
+      "step": 920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.806679070063013e-05,
+      "loss": 0.3474,
+      "step": 921
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8061902627853623e-05,
+      "loss": 0.292,
+      "step": 922
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8057008809953183e-05,
+      "loss": 0.2962,
+      "step": 923
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8052109249081297e-05,
+      "loss": 0.3468,
+      "step": 924
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.804720394739297e-05,
+      "loss": 0.2347,
+      "step": 925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8042292907045756e-05,
+      "loss": 0.3078,
+      "step": 926
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8037376130199702e-05,
+      "loss": 0.31,
+      "step": 927
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8032453619017405e-05,
+      "loss": 0.2748,
+      "step": 928
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8027525375663978e-05,
+      "loss": 0.2524,
+      "step": 929
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.802259140230704e-05,
+      "loss": 0.2947,
+      "step": 930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8017651701116737e-05,
+      "loss": 0.3538,
+      "step": 931
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8012706274265755e-05,
+      "loss": 0.2762,
+      "step": 932
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.800775512392927e-05,
+      "loss": 0.3592,
+      "step": 933
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.800279825228499e-05,
+      "loss": 0.3178,
+      "step": 934
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7997835661513135e-05,
+      "loss": 0.2796,
+      "step": 935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7992867353796447e-05,
+      "loss": 0.3407,
+      "step": 936
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7987893331320174e-05,
+      "loss": 0.3069,
+      "step": 937
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7982913596272077e-05,
+      "loss": 0.3026,
+      "step": 938
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.797792815084244e-05,
+      "loss": 0.2568,
+      "step": 939
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7972936997224052e-05,
+      "loss": 0.3154,
+      "step": 940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.796794013761221e-05,
+      "loss": 0.241,
+      "step": 941
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7962937574204727e-05,
+      "loss": 0.2869,
+      "step": 942
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.795792930920192e-05,
+      "loss": 0.2578,
+      "step": 943
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7952915344806617e-05,
+      "loss": 0.288,
+      "step": 944
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7947895683224145e-05,
+      "loss": 0.3229,
+      "step": 945
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7942870326662353e-05,
+      "loss": 0.274,
+      "step": 946
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7937839277331572e-05,
+      "loss": 0.3132,
+      "step": 947
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7932802537444665e-05,
+      "loss": 0.2683,
+      "step": 948
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.792776010921697e-05,
+      "loss": 0.3258,
+      "step": 949
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.792271199486635e-05,
+      "loss": 0.2831,
+      "step": 950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7917658196613143e-05,
+      "loss": 0.3254,
+      "step": 951
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7912598716680217e-05,
+      "loss": 0.2985,
+      "step": 952
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7907533557292927e-05,
+      "loss": 0.3213,
+      "step": 953
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7902462720679113e-05,
+      "loss": 0.3267,
+      "step": 954
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.789738620906912e-05,
+      "loss": 0.4156,
+      "step": 955
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7892304024695806e-05,
+      "loss": 0.3201,
+      "step": 956
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7887216169794497e-05,
+      "loss": 0.2789,
+      "step": 957
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7882122646603036e-05,
+      "loss": 0.3123,
+      "step": 958
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7877023457361746e-05,
+      "loss": 0.2435,
+      "step": 959
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7871918604313436e-05,
+      "loss": 0.2532,
+      "step": 960
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7866808089703433e-05,
+      "loss": 0.2998,
+      "step": 961
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.786169191577952e-05,
+      "loss": 0.3115,
+      "step": 962
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7856570084791998e-05,
+      "loss": 0.3153,
+      "step": 963
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7851442598993634e-05,
+      "loss": 0.2529,
+      "step": 964
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7846309460639698e-05,
+      "loss": 0.2513,
+      "step": 965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7841170671987942e-05,
+      "loss": 0.3403,
+      "step": 966
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7836026235298594e-05,
+      "loss": 0.288,
+      "step": 967
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7830876152834382e-05,
+      "loss": 0.2719,
+      "step": 968
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7825720426860504e-05,
+      "loss": 0.2876,
+      "step": 969
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.782055905964465e-05,
+      "loss": 0.3011,
+      "step": 970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.781539205345698e-05,
+      "loss": 0.292,
+      "step": 971
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7810219410570146e-05,
+      "loss": 0.3089,
+      "step": 972
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7805041133259273e-05,
+      "loss": 0.278,
+      "step": 973
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7799857223801968e-05,
+      "loss": 0.3021,
+      "step": 974
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7794667684478305e-05,
+      "loss": 0.2573,
+      "step": 975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.778947251757085e-05,
+      "loss": 0.3033,
+      "step": 976
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7784271725364628e-05,
+      "loss": 0.3256,
+      "step": 977
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7779065310147152e-05,
+      "loss": 0.3183,
+      "step": 978
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.77738532742084e-05,
+      "loss": 0.3452,
+      "step": 979
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7768635619840824e-05,
+      "loss": 0.3492,
+      "step": 980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.776341234933935e-05,
+      "loss": 0.3488,
+      "step": 981
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.775818346500137e-05,
+      "loss": 0.3394,
+      "step": 982
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7752948969126747e-05,
+      "loss": 0.2692,
+      "step": 983
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7747708864017815e-05,
+      "loss": 0.2624,
+      "step": 984
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.774246315197937e-05,
+      "loss": 0.2371,
+      "step": 985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.773721183531868e-05,
+      "loss": 0.3015,
+      "step": 986
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.773195491634547e-05,
+      "loss": 0.2647,
+      "step": 987
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7726692397371936e-05,
+      "loss": 0.2981,
+      "step": 988
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7721424280712734e-05,
+      "loss": 0.3399,
+      "step": 989
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7716150568684985e-05,
+      "loss": 0.3128,
+      "step": 990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7710871263608268e-05,
+      "loss": 0.3394,
+      "step": 991
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7705586367804626e-05,
+      "loss": 0.3133,
+      "step": 992
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.770029588359855e-05,
+      "loss": 0.341,
+      "step": 993
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7694999813317006e-05,
+      "loss": 0.3468,
+      "step": 994
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7689698159289402e-05,
+      "loss": 0.345,
+      "step": 995
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.768439092384761e-05,
+      "loss": 0.2523,
+      "step": 996
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7679078109325957e-05,
+      "loss": 0.2427,
+      "step": 997
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.767375971806122e-05,
+      "loss": 0.3109,
+      "step": 998
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7668435752392635e-05,
+      "loss": 0.2437,
+      "step": 999
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.766310621466188e-05,
+      "loss": 0.2927,
+      "step": 1000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.765777110721309e-05,
+      "loss": 0.323,
+      "step": 1001
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7652430432392845e-05,
+      "loss": 0.3537,
+      "step": 1002
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7647084192550187e-05,
+      "loss": 0.3329,
+      "step": 1003
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7641732390036596e-05,
+      "loss": 0.3079,
+      "step": 1004
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.763637502720599e-05,
+      "loss": 0.2925,
+      "step": 1005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.763101210641475e-05,
+      "loss": 0.3075,
+      "step": 1006
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7625643630021696e-05,
+      "loss": 0.3047,
+      "step": 1007
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7620269600388082e-05,
+      "loss": 0.2991,
+      "step": 1008
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7614890019877608e-05,
+      "loss": 0.2435,
+      "step": 1009
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7609504890856434e-05,
+      "loss": 0.3455,
+      "step": 1010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.760411421569313e-05,
+      "loss": 0.2377,
+      "step": 1011
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.759871799675873e-05,
+      "loss": 0.2895,
+      "step": 1012
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7593316236426695e-05,
+      "loss": 0.3389,
+      "step": 1013
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7587908937072925e-05,
+      "loss": 0.3024,
+      "step": 1014
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.758249610107576e-05,
+      "loss": 0.309,
+      "step": 1015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7577077730815966e-05,
+      "loss": 0.286,
+      "step": 1016
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.757165382867675e-05,
+      "loss": 0.2943,
+      "step": 1017
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7566224397043763e-05,
+      "loss": 0.2383,
+      "step": 1018
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.756078943830506e-05,
+      "loss": 0.2373,
+      "step": 1019
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7555348954851153e-05,
+      "loss": 0.2372,
+      "step": 1020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7549902949074973e-05,
+      "loss": 0.3527,
+      "step": 1021
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7544451423371885e-05,
+      "loss": 0.3416,
+      "step": 1022
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.753899438013967e-05,
+      "loss": 0.2334,
+      "step": 1023
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7533531821778553e-05,
+      "loss": 0.2915,
+      "step": 1024
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7528063750691166e-05,
+      "loss": 0.3274,
+      "step": 1025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7522590169282586e-05,
+      "loss": 0.2554,
+      "step": 1026
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7517111079960298e-05,
+      "loss": 0.2741,
+      "step": 1027
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7511626485134214e-05,
+      "loss": 0.2988,
+      "step": 1028
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.750613638721667e-05,
+      "loss": 0.2908,
+      "step": 1029
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7500640788622424e-05,
+      "loss": 0.3102,
+      "step": 1030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7495139691768644e-05,
+      "loss": 0.3551,
+      "step": 1031
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7489633099074922e-05,
+      "loss": 0.2805,
+      "step": 1032
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7484121012963278e-05,
+      "loss": 0.3232,
+      "step": 1033
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7478603435858126e-05,
+      "loss": 0.3173,
+      "step": 1034
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.747308037018632e-05,
+      "loss": 0.3172,
+      "step": 1035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7467551818377105e-05,
+      "loss": 0.3082,
+      "step": 1036
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.746201778286215e-05,
+      "loss": 0.3126,
+      "step": 1037
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.745647826607554e-05,
+      "loss": 0.3066,
+      "step": 1038
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7450933270453766e-05,
+      "loss": 0.2843,
+      "step": 1039
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7445382798435727e-05,
+      "loss": 0.2383,
+      "step": 1040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.743982685246273e-05,
+      "loss": 0.2933,
+      "step": 1041
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7434265434978493e-05,
+      "loss": 0.2692,
+      "step": 1042
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7428698548429146e-05,
+      "loss": 0.2992,
+      "step": 1043
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7423126195263215e-05,
+      "loss": 0.3306,
+      "step": 1044
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7417548377931628e-05,
+      "loss": 0.2837,
+      "step": 1045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.741196509888773e-05,
+      "loss": 0.2903,
+      "step": 1046
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7406376360587262e-05,
+      "loss": 0.2814,
+      "step": 1047
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.740078216548836e-05,
+      "loss": 0.2931,
+      "step": 1048
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7395182516051566e-05,
+      "loss": 0.2947,
+      "step": 1049
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7389577414739816e-05,
+      "loss": 0.2603,
+      "step": 1050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.738396686401845e-05,
+      "loss": 0.3566,
+      "step": 1051
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7378350866355207e-05,
+      "loss": 0.2559,
+      "step": 1052
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7372729424220215e-05,
+      "loss": 0.238,
+      "step": 1053
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7367102540085998e-05,
+      "loss": 0.2839,
+      "step": 1054
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7361470216427472e-05,
+      "loss": 0.2178,
+      "step": 1055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7355832455721955e-05,
+      "loss": 0.285,
+      "step": 1056
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7350189260449137e-05,
+      "loss": 0.2103,
+      "step": 1057
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.734454063309113e-05,
+      "loss": 0.2989,
+      "step": 1058
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7338886576132402e-05,
+      "loss": 0.3343,
+      "step": 1059
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7333227092059825e-05,
+      "loss": 0.2492,
+      "step": 1060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7327562183362662e-05,
+      "loss": 0.3026,
+      "step": 1061
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.732189185253255e-05,
+      "loss": 0.2731,
+      "step": 1062
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.731621610206352e-05,
+      "loss": 0.2424,
+      "step": 1063
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7310534934451984e-05,
+      "loss": 0.3476,
+      "step": 1064
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7304848352196745e-05,
+      "loss": 0.3446,
+      "step": 1065
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.729915635779896e-05,
+      "loss": 0.2678,
+      "step": 1066
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.72934589537622e-05,
+      "loss": 0.2905,
+      "step": 1067
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7287756142592396e-05,
+      "loss": 0.2756,
+      "step": 1068
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7282047926797863e-05,
+      "loss": 0.3107,
+      "step": 1069
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.72763343088893e-05,
+      "loss": 0.3352,
+      "step": 1070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.727061529137976e-05,
+      "loss": 0.237,
+      "step": 1071
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7264890876784698e-05,
+      "loss": 0.3244,
+      "step": 1072
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.725916106762192e-05,
+      "loss": 0.2598,
+      "step": 1073
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7253425866411625e-05,
+      "loss": 0.2839,
+      "step": 1074
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.724768527567637e-05,
+      "loss": 0.2334,
+      "step": 1075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.724193929794108e-05,
+      "loss": 0.308,
+      "step": 1076
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7236187935733065e-05,
+      "loss": 0.3583,
+      "step": 1077
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.723043119158199e-05,
+      "loss": 0.317,
+      "step": 1078
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.722466906801989e-05,
+      "loss": 0.2192,
+      "step": 1079
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7218901567581166e-05,
+      "loss": 0.3163,
+      "step": 1080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7213128692802588e-05,
+      "loss": 0.2966,
+      "step": 1081
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.720735044622329e-05,
+      "loss": 0.2179,
+      "step": 1082
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.720156683038476e-05,
+      "loss": 0.3806,
+      "step": 1083
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.719577784783085e-05,
+      "loss": 0.2907,
+      "step": 1084
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7189983501107784e-05,
+      "loss": 0.249,
+      "step": 1085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7184183792764132e-05,
+      "loss": 0.3126,
+      "step": 1086
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.717837872535083e-05,
+      "loss": 0.2857,
+      "step": 1087
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7172568301421174e-05,
+      "loss": 0.2544,
+      "step": 1088
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7166752523530794e-05,
+      "loss": 0.2837,
+      "step": 1089
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.716093139423771e-05,
+      "loss": 0.249,
+      "step": 1090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7155104916102268e-05,
+      "loss": 0.3106,
+      "step": 1091
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7149273091687173e-05,
+      "loss": 0.2622,
+      "step": 1092
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7143435923557486e-05,
+      "loss": 0.2602,
+      "step": 1093
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7137593414280623e-05,
+      "loss": 0.3154,
+      "step": 1094
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7131745566426334e-05,
+      "loss": 0.2548,
+      "step": 1095
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7125892382566732e-05,
+      "loss": 0.3017,
+      "step": 1096
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7120033865276266e-05,
+      "loss": 0.3398,
+      "step": 1097
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7114170017131746e-05,
+      "loss": 0.2261,
+      "step": 1098
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7108300840712305e-05,
+      "loss": 0.2819,
+      "step": 1099
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7102426338599433e-05,
+      "loss": 0.2752,
+      "step": 1100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.709654651337697e-05,
+      "loss": 0.2215,
+      "step": 1101
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.709066136763107e-05,
+      "loss": 0.2575,
+      "step": 1102
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7084770903950265e-05,
+      "loss": 0.3053,
+      "step": 1103
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7078875124925394e-05,
+      "loss": 0.2909,
+      "step": 1104
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7072974033149646e-05,
+      "loss": 0.2746,
+      "step": 1105
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7067067631218554e-05,
+      "loss": 0.2752,
+      "step": 1106
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7061155921729972e-05,
+      "loss": 0.2443,
+      "step": 1107
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7055238907284097e-05,
+      "loss": 0.3122,
+      "step": 1108
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.704931659048346e-05,
+      "loss": 0.3468,
+      "step": 1109
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7043388973932923e-05,
+      "loss": 0.2897,
+      "step": 1110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.703745606023968e-05,
+      "loss": 0.2598,
+      "step": 1111
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7031517852013247e-05,
+      "loss": 0.2957,
+      "step": 1112
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7025574351865475e-05,
+      "loss": 0.3186,
+      "step": 1113
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7019625562410547e-05,
+      "loss": 0.2639,
+      "step": 1114
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.701367148626497e-05,
+      "loss": 0.2605,
+      "step": 1115
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7007712126047568e-05,
+      "loss": 0.2358,
+      "step": 1116
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7001747484379502e-05,
+      "loss": 0.2337,
+      "step": 1117
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.699577756388424e-05,
+      "loss": 0.3448,
+      "step": 1118
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.698980236718759e-05,
+      "loss": 0.2796,
+      "step": 1119
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6983821896917663e-05,
+      "loss": 0.2625,
+      "step": 1120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.697783615570491e-05,
+      "loss": 0.3393,
+      "step": 1121
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.697184514618208e-05,
+      "loss": 0.3139,
+      "step": 1122
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6965848870984254e-05,
+      "loss": 0.2897,
+      "step": 1123
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6959847332748814e-05,
+      "loss": 0.2363,
+      "step": 1124
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6953840534115475e-05,
+      "loss": 0.228,
+      "step": 1125
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6947828477726253e-05,
+      "loss": 0.3199,
+      "step": 1126
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.694181116622548e-05,
+      "loss": 0.3281,
+      "step": 1127
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.69357886022598e-05,
+      "loss": 0.2885,
+      "step": 1128
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.692976078847816e-05,
+      "loss": 0.3191,
+      "step": 1129
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6923727727531836e-05,
+      "loss": 0.2121,
+      "step": 1130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.691768942207439e-05,
+      "loss": 0.2771,
+      "step": 1131
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.69116458747617e-05,
+      "loss": 0.2873,
+      "step": 1132
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6905597088251957e-05,
+      "loss": 0.2849,
+      "step": 1133
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6899543065205633e-05,
+      "loss": 0.2694,
+      "step": 1134
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6893483808285536e-05,
+      "loss": 0.2937,
+      "step": 1135
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6887419320156748e-05,
+      "loss": 0.2586,
+      "step": 1136
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6881349603486662e-05,
+      "loss": 0.1737,
+      "step": 1137
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6875274660944978e-05,
+      "loss": 0.3164,
+      "step": 1138
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.686919449520368e-05,
+      "loss": 0.2919,
+      "step": 1139
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6863109108937065e-05,
+      "loss": 0.3105,
+      "step": 1140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6857018504821716e-05,
+      "loss": 0.3236,
+      "step": 1141
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.685092268553651e-05,
+      "loss": 0.2749,
+      "step": 1142
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6844821653762622e-05,
+      "loss": 0.2846,
+      "step": 1143
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6838715412183524e-05,
+      "loss": 0.2667,
+      "step": 1144
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.683260396348497e-05,
+      "loss": 0.2451,
+      "step": 1145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.682648731035501e-05,
+      "loss": 0.2757,
+      "step": 1146
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.682036545548398e-05,
+      "loss": 0.2481,
+      "step": 1147
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.681423840156451e-05,
+      "loss": 0.2129,
+      "step": 1148
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.680810615129151e-05,
+      "loss": 0.2586,
+      "step": 1149
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6801968707362172e-05,
+      "loss": 0.2526,
+      "step": 1150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.679582607247599e-05,
+      "loss": 0.3058,
+      "step": 1151
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6789678249334724e-05,
+      "loss": 0.2549,
+      "step": 1152
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6783525240642416e-05,
+      "loss": 0.2801,
+      "step": 1153
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6777367049105405e-05,
+      "loss": 0.3158,
+      "step": 1154
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.677120367743229e-05,
+      "loss": 0.2323,
+      "step": 1155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6765035128333963e-05,
+      "loss": 0.308,
+      "step": 1156
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.675886140452358e-05,
+      "loss": 0.2254,
+      "step": 1157
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6752682508716593e-05,
+      "loss": 0.2978,
+      "step": 1158
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6746498443630708e-05,
+      "loss": 0.2745,
+      "step": 1159
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6740309211985916e-05,
+      "loss": 0.2219,
+      "step": 1160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6734114816504478e-05,
+      "loss": 0.2314,
+      "step": 1161
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6727915259910918e-05,
+      "loss": 0.2913,
+      "step": 1162
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.672171054493205e-05,
+      "loss": 0.2516,
+      "step": 1163
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6715500674296938e-05,
+      "loss": 0.2563,
+      "step": 1164
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.670928565073692e-05,
+      "loss": 0.3334,
+      "step": 1165
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.67030654769856e-05,
+      "loss": 0.2667,
+      "step": 1166
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6696840155778854e-05,
+      "loss": 0.2615,
+      "step": 1167
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6690609689854813e-05,
+      "loss": 0.2596,
+      "step": 1168
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6684374081953874e-05,
+      "loss": 0.2618,
+      "step": 1169
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6678133334818688e-05,
+      "loss": 0.2686,
+      "step": 1170
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.667188745119419e-05,
+      "loss": 0.2772,
+      "step": 1171
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6665636433827547e-05,
+      "loss": 0.2949,
+      "step": 1172
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.66593802854682e-05,
+      "loss": 0.2969,
+      "step": 1173
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6653119008867835e-05,
+      "loss": 0.2982,
+      "step": 1174
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6646852606780406e-05,
+      "loss": 0.2754,
+      "step": 1175
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6640581081962118e-05,
+      "loss": 0.2398,
+      "step": 1176
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6634304437171425e-05,
+      "loss": 0.2869,
+      "step": 1177
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6628022675169034e-05,
+      "loss": 0.305,
+      "step": 1178
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6621735798717912e-05,
+      "loss": 0.3015,
+      "step": 1179
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6615443810583253e-05,
+      "loss": 0.2678,
+      "step": 1180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6609146713532523e-05,
+      "loss": 0.2651,
+      "step": 1181
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6602844510335432e-05,
+      "loss": 0.1787,
+      "step": 1182
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6596537203763916e-05,
+      "loss": 0.2866,
+      "step": 1183
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6590224796592177e-05,
+      "loss": 0.2277,
+      "step": 1184
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.658390729159665e-05,
+      "loss": 0.3123,
+      "step": 1185
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.657758469155602e-05,
+      "loss": 0.252,
+      "step": 1186
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6571256999251206e-05,
+      "loss": 0.3159,
+      "step": 1187
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6564924217465365e-05,
+      "loss": 0.199,
+      "step": 1188
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.65585863489839e-05,
+      "loss": 0.2698,
+      "step": 1189
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6552243396594447e-05,
+      "loss": 0.2866,
+      "step": 1190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.654589536308688e-05,
+      "loss": 0.2677,
+      "step": 1191
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6539542251253303e-05,
+      "loss": 0.2877,
+      "step": 1192
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6533184063888065e-05,
+      "loss": 0.2855,
+      "step": 1193
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.652682080378773e-05,
+      "loss": 0.2514,
+      "step": 1194
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.652045247375111e-05,
+      "loss": 0.2711,
+      "step": 1195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.651407907657924e-05,
+      "loss": 0.2389,
+      "step": 1196
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.650770061507538e-05,
+      "loss": 0.2192,
+      "step": 1197
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.650131709204502e-05,
+      "loss": 0.296,
+      "step": 1198
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6494928510295883e-05,
+      "loss": 0.251,
+      "step": 1199
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6488534872637908e-05,
+      "loss": 0.2574,
+      "step": 1200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6482136181883263e-05,
+      "loss": 0.2579,
+      "step": 1201
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.647573244084633e-05,
+      "loss": 0.2921,
+      "step": 1202
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.646932365234373e-05,
+      "loss": 0.2716,
+      "step": 1203
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6462909819194283e-05,
+      "loss": 0.2966,
+      "step": 1204
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6456490944219044e-05,
+      "loss": 0.27,
+      "step": 1205
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.645006703024128e-05,
+      "loss": 0.3286,
+      "step": 1206
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6443638080086466e-05,
+      "loss": 0.2567,
+      "step": 1207
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6437204096582314e-05,
+      "loss": 0.2771,
+      "step": 1208
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.643076508255872e-05,
+      "loss": 0.3063,
+      "step": 1209
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.642432104084782e-05,
+      "loss": 0.3114,
+      "step": 1210
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.641787197428394e-05,
+      "loss": 0.2347,
+      "step": 1211
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6411417885703636e-05,
+      "loss": 0.3002,
+      "step": 1212
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.640495877794566e-05,
+      "loss": 0.3109,
+      "step": 1213
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.639849465385097e-05,
+      "loss": 0.3202,
+      "step": 1214
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6392025516262742e-05,
+      "loss": 0.2495,
+      "step": 1215
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.638555136802634e-05,
+      "loss": 0.267,
+      "step": 1216
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6379072211989348e-05,
+      "loss": 0.2216,
+      "step": 1217
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6372588051001547e-05,
+      "loss": 0.2774,
+      "step": 1218
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6366098887914917e-05,
+      "loss": 0.2606,
+      "step": 1219
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6359604725583644e-05,
+      "loss": 0.3108,
+      "step": 1220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.63531055668641e-05,
+      "loss": 0.2481,
+      "step": 1221
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.634660141461487e-05,
+      "loss": 0.242,
+      "step": 1222
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.634009227169673e-05,
+      "loss": 0.2254,
+      "step": 1223
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6333578140972647e-05,
+      "loss": 0.2026,
+      "step": 1224
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.632705902530778e-05,
+      "loss": 0.2917,
+      "step": 1225
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6320534927569493e-05,
+      "loss": 0.3095,
+      "step": 1226
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.631400585062733e-05,
+      "loss": 0.2686,
+      "step": 1227
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6307471797353034e-05,
+      "loss": 0.2494,
+      "step": 1228
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6300932770620518e-05,
+      "loss": 0.2505,
+      "step": 1229
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.629438877330591e-05,
+      "loss": 0.3199,
+      "step": 1230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.62878398082875e-05,
+      "loss": 0.2523,
+      "step": 1231
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6281285878445778e-05,
+      "loss": 0.3043,
+      "step": 1232
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.627472698666342e-05,
+      "loss": 0.268,
+      "step": 1233
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.626816313582526e-05,
+      "loss": 0.2598,
+      "step": 1234
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6261594328818346e-05,
+      "loss": 0.2546,
+      "step": 1235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6255020568531882e-05,
+      "loss": 0.2828,
+      "step": 1236
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6248441857857256e-05,
+      "loss": 0.29,
+      "step": 1237
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.624185819968805e-05,
+      "loss": 0.2373,
+      "step": 1238
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6235269596919996e-05,
+      "loss": 0.2851,
+      "step": 1239
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6228676052451017e-05,
+      "loss": 0.2277,
+      "step": 1240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.622207756918121e-05,
+      "loss": 0.2621,
+      "step": 1241
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.621547415001283e-05,
+      "loss": 0.2729,
+      "step": 1242
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.620886579785032e-05,
+      "loss": 0.2941,
+      "step": 1243
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.620225251560029e-05,
+      "loss": 0.2626,
+      "step": 1244
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6195634306171507e-05,
+      "loss": 0.25,
+      "step": 1245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.618901117247491e-05,
+      "loss": 0.2918,
+      "step": 1246
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.618238311742362e-05,
+      "loss": 0.3457,
+      "step": 1247
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6175750143932898e-05,
+      "loss": 0.2543,
+      "step": 1248
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6169112254920176e-05,
+      "loss": 0.273,
+      "step": 1249
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.616246945330506e-05,
+      "loss": 0.2411,
+      "step": 1250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6155821742009304e-05,
+      "loss": 0.2359,
+      "step": 1251
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6149169123956826e-05,
+      "loss": 0.2505,
+      "step": 1252
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6142511602073705e-05,
+      "loss": 0.2888,
+      "step": 1253
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6135849179288166e-05,
+      "loss": 0.2247,
+      "step": 1254
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.61291818585306e-05,
+      "loss": 0.2746,
+      "step": 1255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6122509642733554e-05,
+      "loss": 0.2004,
+      "step": 1256
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6115832534831716e-05,
+      "loss": 0.2476,
+      "step": 1257
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6109150537761947e-05,
+      "loss": 0.2864,
+      "step": 1258
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6102463654463227e-05,
+      "loss": 0.2867,
+      "step": 1259
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6095771887876712e-05,
+      "loss": 0.2283,
+      "step": 1260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6089075240945697e-05,
+      "loss": 0.282,
+      "step": 1261
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.608237371661562e-05,
+      "loss": 0.3375,
+      "step": 1262
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.607566731783407e-05,
+      "loss": 0.2545,
+      "step": 1263
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.606895604755078e-05,
+      "loss": 0.275,
+      "step": 1264
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.606223990871762e-05,
+      "loss": 0.3054,
+      "step": 1265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.60555189042886e-05,
+      "loss": 0.1799,
+      "step": 1266
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6048793037219883e-05,
+      "loss": 0.2665,
+      "step": 1267
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6042062310469754e-05,
+      "loss": 0.2643,
+      "step": 1268
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6035326726998654e-05,
+      "loss": 0.2473,
+      "step": 1269
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.602858628976914e-05,
+      "loss": 0.3047,
+      "step": 1270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6021841001745915e-05,
+      "loss": 0.2534,
+      "step": 1271
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.601509086589583e-05,
+      "loss": 0.3199,
+      "step": 1272
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6008335885187826e-05,
+      "loss": 0.2365,
+      "step": 1273
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6001576062593024e-05,
+      "loss": 0.2546,
+      "step": 1274
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5994811401084643e-05,
+      "loss": 0.2636,
+      "step": 1275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5988041903638043e-05,
+      "loss": 0.2402,
+      "step": 1276
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5981267573230704e-05,
+      "loss": 0.2804,
+      "step": 1277
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5974488412842233e-05,
+      "loss": 0.237,
+      "step": 1278
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5967704425454373e-05,
+      "loss": 0.2732,
+      "step": 1279
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.596091561405097e-05,
+      "loss": 0.2909,
+      "step": 1280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5954121981618016e-05,
+      "loss": 0.266,
+      "step": 1281
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5947323531143598e-05,
+      "loss": 0.2746,
+      "step": 1282
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5940520265617943e-05,
+      "loss": 0.2345,
+      "step": 1283
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5933712188033377e-05,
+      "loss": 0.2279,
+      "step": 1284
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5926899301384363e-05,
+      "loss": 0.2309,
+      "step": 1285
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.592008160866746e-05,
+      "loss": 0.2888,
+      "step": 1286
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.591325911288136e-05,
+      "loss": 0.2142,
+      "step": 1287
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5906431817026847e-05,
+      "loss": 0.2204,
+      "step": 1288
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5899599724106836e-05,
+      "loss": 0.2498,
+      "step": 1289
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5892762837126334e-05,
+      "loss": 0.2405,
+      "step": 1290
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5885921159092472e-05,
+      "loss": 0.2632,
+      "step": 1291
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.587907469301448e-05,
+      "loss": 0.2034,
+      "step": 1292
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5872223441903688e-05,
+      "loss": 0.3207,
+      "step": 1293
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5865367408773548e-05,
+      "loss": 0.2693,
+      "step": 1294
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5858506596639603e-05,
+      "loss": 0.2541,
+      "step": 1295
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5851641008519495e-05,
+      "loss": 0.2258,
+      "step": 1296
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.584477064743298e-05,
+      "loss": 0.2628,
+      "step": 1297
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.58378955164019e-05,
+      "loss": 0.2167,
+      "step": 1298
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.58310156184502e-05,
+      "loss": 0.1392,
+      "step": 1299
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5824130956603924e-05,
+      "loss": 0.2332,
+      "step": 1300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.581724153389121e-05,
+      "loss": 0.1885,
+      "step": 1301
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5810347353342285e-05,
+      "loss": 0.2019,
+      "step": 1302
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.580344841798948e-05,
+      "loss": 0.2766,
+      "step": 1303
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5796544730867206e-05,
+      "loss": 0.2326,
+      "step": 1304
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.578963629501197e-05,
+      "loss": 0.2078,
+      "step": 1305
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5782723113462366e-05,
+      "loss": 0.2807,
+      "step": 1306
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5775805189259075e-05,
+      "loss": 0.2763,
+      "step": 1307
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.576888252544486e-05,
+      "loss": 0.2942,
+      "step": 1308
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.576195512506458e-05,
+      "loss": 0.2676,
+      "step": 1309
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5755022991165168e-05,
+      "loss": 0.2472,
+      "step": 1310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.574808612679564e-05,
+      "loss": 0.2329,
+      "step": 1311
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5741144535007097e-05,
+      "loss": 0.2923,
+      "step": 1312
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5734198218852717e-05,
+      "loss": 0.2504,
+      "step": 1313
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5727247181387747e-05,
+      "loss": 0.2846,
+      "step": 1314
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.572029142566953e-05,
+      "loss": 0.2548,
+      "step": 1315
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5713330954757458e-05,
+      "loss": 0.2542,
+      "step": 1316
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5706365771713032e-05,
+      "loss": 0.2338,
+      "step": 1317
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5699395879599788e-05,
+      "loss": 0.2619,
+      "step": 1318
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5692421281483366e-05,
+      "loss": 0.3352,
+      "step": 1319
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.568544198043145e-05,
+      "loss": 0.214,
+      "step": 1320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.567845797951381e-05,
+      "loss": 0.3434,
+      "step": 1321
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.567146928180227e-05,
+      "loss": 0.3014,
+      "step": 1322
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.566447589037073e-05,
+      "loss": 0.2515,
+      "step": 1323
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5657477808295152e-05,
+      "loss": 0.2755,
+      "step": 1324
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5650475038653555e-05,
+      "loss": 0.297,
+      "step": 1325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5643467584526028e-05,
+      "loss": 0.2613,
+      "step": 1326
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.563645544899472e-05,
+      "loss": 0.2875,
+      "step": 1327
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5629438635143833e-05,
+      "loss": 0.2447,
+      "step": 1328
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.562241714605963e-05,
+      "loss": 0.2548,
+      "step": 1329
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.561539098483043e-05,
+      "loss": 0.2418,
+      "step": 1330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5608360154546607e-05,
+      "loss": 0.2579,
+      "step": 1331
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.560132465830059e-05,
+      "loss": 0.2961,
+      "step": 1332
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5594284499186856e-05,
+      "loss": 0.2188,
+      "step": 1333
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.558723968030194e-05,
+      "loss": 0.2944,
+      "step": 1334
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5580190204744423e-05,
+      "loss": 0.2224,
+      "step": 1335
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5573136075614932e-05,
+      "loss": 0.2443,
+      "step": 1336
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5566077296016137e-05,
+      "loss": 0.2766,
+      "step": 1337
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.555901386905277e-05,
+      "loss": 0.2382,
+      "step": 1338
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5551945797831587e-05,
+      "loss": 0.2815,
+      "step": 1339
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.55448730854614e-05,
+      "loss": 0.2734,
+      "step": 1340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.553779573505306e-05,
+      "loss": 0.2218,
+      "step": 1341
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5530713749719455e-05,
+      "loss": 0.2571,
+      "step": 1342
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5523627132575504e-05,
+      "loss": 0.2305,
+      "step": 1343
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5516535886738187e-05,
+      "loss": 0.2094,
+      "step": 1344
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5509440015326498e-05,
+      "loss": 0.2767,
+      "step": 1345
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5502339521461466e-05,
+      "loss": 0.1902,
+      "step": 1346
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.549523440826617e-05,
+      "loss": 0.2402,
+      "step": 1347
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5488124678865702e-05,
+      "loss": 0.2308,
+      "step": 1348
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5481010336387198e-05,
+      "loss": 0.2494,
+      "step": 1349
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5473891383959817e-05,
+      "loss": 0.2871,
+      "step": 1350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.546676782471474e-05,
+      "loss": 0.3133,
+      "step": 1351
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5459639661785194e-05,
+      "loss": 0.2609,
+      "step": 1352
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.545250689830641e-05,
+      "loss": 0.2777,
+      "step": 1353
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5445369537415643e-05,
+      "loss": 0.2273,
+      "step": 1354
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5438227582252186e-05,
+      "loss": 0.2237,
+      "step": 1355
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5431081035957344e-05,
+      "loss": 0.2525,
+      "step": 1356
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5423929901674434e-05,
+      "loss": 0.2217,
+      "step": 1357
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5416774182548806e-05,
+      "loss": 0.2585,
+      "step": 1358
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5409613881727817e-05,
+      "loss": 0.2162,
+      "step": 1359
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5402449002360837e-05,
+      "loss": 0.2773,
+      "step": 1360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5395279547599255e-05,
+      "loss": 0.2226,
+      "step": 1361
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.538810552059647e-05,
+      "loss": 0.2865,
+      "step": 1362
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.53809269245079e-05,
+      "loss": 0.2515,
+      "step": 1363
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.537374376249096e-05,
+      "loss": 0.2048,
+      "step": 1364
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5366556037705077e-05,
+      "loss": 0.2062,
+      "step": 1365
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5359363753311698e-05,
+      "loss": 0.2234,
+      "step": 1366
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.535216691247425e-05,
+      "loss": 0.2468,
+      "step": 1367
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5344965518358193e-05,
+      "loss": 0.2169,
+      "step": 1368
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5337759574130957e-05,
+      "loss": 0.283,
+      "step": 1369
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.533054908296201e-05,
+      "loss": 0.2602,
+      "step": 1370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5323334048022792e-05,
+      "loss": 0.2428,
+      "step": 1371
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5316114472486757e-05,
+      "loss": 0.2864,
+      "step": 1372
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5308890359529342e-05,
+      "loss": 0.2454,
+      "step": 1373
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5301661712328e-05,
+      "loss": 0.2162,
+      "step": 1374
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5294428534062147e-05,
+      "loss": 0.2506,
+      "step": 1375
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5287190827913232e-05,
+      "loss": 0.3124,
+      "step": 1376
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5279948597064664e-05,
+      "loss": 0.2352,
+      "step": 1377
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.527270184470185e-05,
+      "loss": 0.3348,
+      "step": 1378
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.52654505740122e-05,
+      "loss": 0.264,
+      "step": 1379
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.525819478818509e-05,
+      "loss": 0.2286,
+      "step": 1380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5250934490411895e-05,
+      "loss": 0.2605,
+      "step": 1381
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5243669683885974e-05,
+      "loss": 0.2154,
+      "step": 1382
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.523640037180266e-05,
+      "loss": 0.2296,
+      "step": 1383
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5229126557359284e-05,
+      "loss": 0.2026,
+      "step": 1384
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5221848243755138e-05,
+      "loss": 0.2442,
+      "step": 1385
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5214565434191503e-05,
+      "loss": 0.2771,
+      "step": 1386
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5207278131871644e-05,
+      "loss": 0.2735,
+      "step": 1387
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.519998634000079e-05,
+      "loss": 0.2059,
+      "step": 1388
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5192690061786157e-05,
+      "loss": 0.2589,
+      "step": 1389
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5185389300436912e-05,
+      "loss": 0.2473,
+      "step": 1390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5178084059164226e-05,
+      "loss": 0.1826,
+      "step": 1391
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5170774341181207e-05,
+      "loss": 0.24,
+      "step": 1392
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.516346014970296e-05,
+      "loss": 0.2825,
+      "step": 1393
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5156141487946546e-05,
+      "loss": 0.2605,
+      "step": 1394
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.514881835913099e-05,
+      "loss": 0.2447,
+      "step": 1395
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5141490766477274e-05,
+      "loss": 0.1997,
+      "step": 1396
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5134158713208364e-05,
+      "loss": 0.2191,
+      "step": 1397
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5126822202549178e-05,
+      "loss": 0.2028,
+      "step": 1398
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.511948123772659e-05,
+      "loss": 0.2994,
+      "step": 1399
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5112135821969438e-05,
+      "loss": 0.2357,
+      "step": 1400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5104785958508517e-05,
+      "loss": 0.2635,
+      "step": 1401
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.509743165057657e-05,
+      "loss": 0.2861,
+      "step": 1402
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5090072901408314e-05,
+      "loss": 0.2947,
+      "step": 1403
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5082709714240405e-05,
+      "loss": 0.2291,
+      "step": 1404
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5075342092311455e-05,
+      "loss": 0.1967,
+      "step": 1405
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.506797003886202e-05,
+      "loss": 0.2018,
+      "step": 1406
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5060593557134616e-05,
+      "loss": 0.27,
+      "step": 1407
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5053212650373695e-05,
+      "loss": 0.2505,
+      "step": 1408
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5045827321825673e-05,
+      "loss": 0.2565,
+      "step": 1409
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.503843757473889e-05,
+      "loss": 0.2207,
+      "step": 1410
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5031043412363644e-05,
+      "loss": 0.1911,
+      "step": 1411
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5023644837952167e-05,
+      "loss": 0.2846,
+      "step": 1412
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5016241854758638e-05,
+      "loss": 0.2788,
+      "step": 1413
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.500883446603917e-05,
+      "loss": 0.1482,
+      "step": 1414
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5001422675051816e-05,
+      "loss": 0.2352,
+      "step": 1415
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4994006485056563e-05,
+      "loss": 0.2527,
+      "step": 1416
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4986585899315346e-05,
+      "loss": 0.2474,
+      "step": 1417
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4979160921092003e-05,
+      "loss": 0.1898,
+      "step": 1418
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4971731553652337e-05,
+      "loss": 0.2507,
+      "step": 1419
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4964297800264063e-05,
+      "loss": 0.2647,
+      "step": 1420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4956859664196833e-05,
+      "loss": 0.2804,
+      "step": 1421
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4949417148722216e-05,
+      "loss": 0.2511,
+      "step": 1422
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4941970257113725e-05,
+      "loss": 0.3006,
+      "step": 1423
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4934518992646777e-05,
+      "loss": 0.264,
+      "step": 1424
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4927063358598733e-05,
+      "loss": 0.2469,
+      "step": 1425
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4919603358248857e-05,
+      "loss": 0.2955,
+      "step": 1426
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4912138994878353e-05,
+      "loss": 0.1966,
+      "step": 1427
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.490467027177032e-05,
+      "loss": 0.3092,
+      "step": 1428
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.48971971922098e-05,
+      "loss": 0.2792,
+      "step": 1429
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4889719759483725e-05,
+      "loss": 0.1689,
+      "step": 1430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.488223797688097e-05,
+      "loss": 0.1941,
+      "step": 1431
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4874751847692302e-05,
+      "loss": 0.3138,
+      "step": 1432
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.486726137521041e-05,
+      "loss": 0.2354,
+      "step": 1433
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.485976656272988e-05,
+      "loss": 0.2849,
+      "step": 1434
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4852267413547234e-05,
+      "loss": 0.2732,
+      "step": 1435
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.484476393096087e-05,
+      "loss": 0.2566,
+      "step": 1436
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4837256118271112e-05,
+      "loss": 0.2426,
+      "step": 1437
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4829743978780182e-05,
+      "loss": 0.2237,
+      "step": 1438
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4822227515792207e-05,
+      "loss": 0.261,
+      "step": 1439
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4814706732613214e-05,
+      "loss": 0.2259,
+      "step": 1440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4807181632551134e-05,
+      "loss": 0.1697,
+      "step": 1441
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.479965221891578e-05,
+      "loss": 0.2252,
+      "step": 1442
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4792118495018897e-05,
+      "loss": 0.237,
+      "step": 1443
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.478458046417409e-05,
+      "loss": 0.2232,
+      "step": 1444
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4777038129696876e-05,
+      "loss": 0.2248,
+      "step": 1445
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4769491494904667e-05,
+      "loss": 0.2612,
+      "step": 1446
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4761940563116757e-05,
+      "loss": 0.1643,
+      "step": 1447
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.475438533765433e-05,
+      "loss": 0.301,
+      "step": 1448
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.474682582184047e-05,
+      "loss": 0.2315,
+      "step": 1449
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4739262019000136e-05,
+      "loss": 0.2963,
+      "step": 1450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.473169393246018e-05,
+      "loss": 0.2471,
+      "step": 1451
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4724121565549344e-05,
+      "loss": 0.2072,
+      "step": 1452
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4716544921598227e-05,
+      "loss": 0.2834,
+      "step": 1453
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.470896400393934e-05,
+      "loss": 0.2835,
+      "step": 1454
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4701378815907055e-05,
+      "loss": 0.2149,
+      "step": 1455
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4693789360837625e-05,
+      "loss": 0.2708,
+      "step": 1456
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4686195642069193e-05,
+      "loss": 0.3006,
+      "step": 1457
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.467859766294175e-05,
+      "loss": 0.2663,
+      "step": 1458
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.46709954267972e-05,
+      "loss": 0.2091,
+      "step": 1459
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4663388936979273e-05,
+      "loss": 0.2376,
+      "step": 1460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4655778196833606e-05,
+      "loss": 0.1971,
+      "step": 1461
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4648163209707693e-05,
+      "loss": 0.2389,
+      "step": 1462
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.46405439789509e-05,
+      "loss": 0.2037,
+      "step": 1463
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4632920507914447e-05,
+      "loss": 0.1771,
+      "step": 1464
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.462529279995144e-05,
+      "loss": 0.3004,
+      "step": 1465
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4617660858416816e-05,
+      "loss": 0.3068,
+      "step": 1466
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4610024686667417e-05,
+      "loss": 0.2188,
+      "step": 1467
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4602384288061915e-05,
+      "loss": 0.1915,
+      "step": 1468
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.459473966596084e-05,
+      "loss": 0.2211,
+      "step": 1469
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4587090823726605e-05,
+      "loss": 0.2361,
+      "step": 1470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.457943776472345e-05,
+      "loss": 0.256,
+      "step": 1471
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.457178049231749e-05,
+      "loss": 0.263,
+      "step": 1472
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.456411900987668e-05,
+      "loss": 0.2074,
+      "step": 1473
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4556453320770838e-05,
+      "loss": 0.2096,
+      "step": 1474
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4548783428371627e-05,
+      "loss": 0.3148,
+      "step": 1475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.454110933605256e-05,
+      "loss": 0.2405,
+      "step": 1476
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4533431047188984e-05,
+      "loss": 0.2378,
+      "step": 1477
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4525748565158125e-05,
+      "loss": 0.3319,
+      "step": 1478
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4518061893339014e-05,
+      "loss": 0.319,
+      "step": 1479
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.451037103511255e-05,
+      "loss": 0.1893,
+      "step": 1480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4502675993861464e-05,
+      "loss": 0.266,
+      "step": 1481
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4494976772970334e-05,
+      "loss": 0.2738,
+      "step": 1482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.448727337582557e-05,
+      "loss": 0.188,
+      "step": 1483
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4479565805815415e-05,
+      "loss": 0.2551,
+      "step": 1484
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4471854066329957e-05,
+      "loss": 0.2317,
+      "step": 1485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4464138160761116e-05,
+      "loss": 0.2785,
+      "step": 1486
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.445641809250264e-05,
+      "loss": 0.2647,
+      "step": 1487
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.444869386495011e-05,
+      "loss": 0.253,
+      "step": 1488
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4440965481500933e-05,
+      "loss": 0.2367,
+      "step": 1489
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4433232945554354e-05,
+      "loss": 0.18,
+      "step": 1490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.442549626051143e-05,
+      "loss": 0.1917,
+      "step": 1491
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4417755429775057e-05,
+      "loss": 0.1693,
+      "step": 1492
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.441001045674995e-05,
+      "loss": 0.2485,
+      "step": 1493
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.440226134484264e-05,
+      "loss": 0.2292,
+      "step": 1494
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.439450809746148e-05,
+      "loss": 0.1474,
+      "step": 1495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4386750718016648e-05,
+      "loss": 0.2467,
+      "step": 1496
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4378989209920136e-05,
+      "loss": 0.2302,
+      "step": 1497
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4371223576585753e-05,
+      "loss": 0.2364,
+      "step": 1498
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4363453821429116e-05,
+      "loss": 0.2179,
+      "step": 1499
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4355679947867666e-05,
+      "loss": 0.2375,
+      "step": 1500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4347901959320646e-05,
+      "loss": 0.2304,
+      "step": 1501
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4340119859209117e-05,
+      "loss": 0.2086,
+      "step": 1502
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4332333650955934e-05,
+      "loss": 0.238,
+      "step": 1503
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4324543337985785e-05,
+      "loss": 0.2707,
+      "step": 1504
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4316748923725135e-05,
+      "loss": 0.2388,
+      "step": 1505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4308950411602268e-05,
+      "loss": 0.2473,
+      "step": 1506
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.430114780504727e-05,
+      "loss": 0.2822,
+      "step": 1507
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.429334110749202e-05,
+      "loss": 0.2882,
+      "step": 1508
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4285530322370208e-05,
+      "loss": 0.2543,
+      "step": 1509
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4277715453117315e-05,
+      "loss": 0.2018,
+      "step": 1510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.426989650317062e-05,
+      "loss": 0.1585,
+      "step": 1511
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4262073475969188e-05,
+      "loss": 0.3601,
+      "step": 1512
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.425424637495389e-05,
+      "loss": 0.2605,
+      "step": 1513
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4246415203567394e-05,
+      "loss": 0.2387,
+      "step": 1514
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.423857996525413e-05,
+      "loss": 0.2146,
+      "step": 1515
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4230740663460353e-05,
+      "loss": 0.2889,
+      "step": 1516
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4222897301634074e-05,
+      "loss": 0.1871,
+      "step": 1517
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.421504988322511e-05,
+      "loss": 0.2087,
+      "step": 1518
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4207198411685055e-05,
+      "loss": 0.2186,
+      "step": 1519
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.419934289046728e-05,
+      "loss": 0.2393,
+      "step": 1520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4191483323026955e-05,
+      "loss": 0.2104,
+      "step": 1521
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.418361971282101e-05,
+      "loss": 0.246,
+      "step": 1522
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4175752063308163e-05,
+      "loss": 0.1849,
+      "step": 1523
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4167880377948904e-05,
+      "loss": 0.2288,
+      "step": 1524
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.416000466020551e-05,
+      "loss": 0.3299,
+      "step": 1525
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4152124913542016e-05,
+      "loss": 0.2188,
+      "step": 1526
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4144241141424233e-05,
+      "loss": 0.2195,
+      "step": 1527
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.413635334731975e-05,
+      "loss": 0.2739,
+      "step": 1528
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.412846153469792e-05,
+      "loss": 0.2638,
+      "step": 1529
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.412056570702986e-05,
+      "loss": 0.2525,
+      "step": 1530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.411266586778846e-05,
+      "loss": 0.2239,
+      "step": 1531
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.410476202044837e-05,
+      "loss": 0.2467,
+      "step": 1532
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.409685416848601e-05,
+      "loss": 0.2349,
+      "step": 1533
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.408894231537955e-05,
+      "loss": 0.2652,
+      "step": 1534
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.408102646460892e-05,
+      "loss": 0.2484,
+      "step": 1535
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4073106619655824e-05,
+      "loss": 0.2258,
+      "step": 1536
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4065182784003706e-05,
+      "loss": 0.2209,
+      "step": 1537
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.405725496113777e-05,
+      "loss": 0.2167,
+      "step": 1538
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4049323154544986e-05,
+      "loss": 0.2342,
+      "step": 1539
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4041387367714056e-05,
+      "loss": 0.2663,
+      "step": 1540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4033447604135443e-05,
+      "loss": 0.2269,
+      "step": 1541
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4025503867301362e-05,
+      "loss": 0.332,
+      "step": 1542
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.401755616070577e-05,
+      "loss": 0.2361,
+      "step": 1543
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.400960448784437e-05,
+      "loss": 0.2276,
+      "step": 1544
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.400164885221462e-05,
+      "loss": 0.1947,
+      "step": 1545
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3993689257315697e-05,
+      "loss": 0.2979,
+      "step": 1546
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3985725706648554e-05,
+      "loss": 0.2277,
+      "step": 1547
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3977758203715853e-05,
+      "loss": 0.2294,
+      "step": 1548
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.396978675202201e-05,
+      "loss": 0.2287,
+      "step": 1549
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3961811355073167e-05,
+      "loss": 0.192,
+      "step": 1550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3953832016377218e-05,
+      "loss": 0.1854,
+      "step": 1551
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.394584873944378e-05,
+      "loss": 0.2478,
+      "step": 1552
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3937861527784197e-05,
+      "loss": 0.3207,
+      "step": 1553
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3929870384911555e-05,
+      "loss": 0.2385,
+      "step": 1554
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3921875314340663e-05,
+      "loss": 0.2649,
+      "step": 1555
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3913876319588058e-05,
+      "loss": 0.2366,
+      "step": 1556
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3905873404172007e-05,
+      "loss": 0.2656,
+      "step": 1557
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3897866571612487e-05,
+      "loss": 0.2888,
+      "step": 1558
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3889855825431225e-05,
+      "loss": 0.2907,
+      "step": 1559
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3881841169151637e-05,
+      "loss": 0.198,
+      "step": 1560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3873822606298888e-05,
+      "loss": 0.2369,
+      "step": 1561
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3865800140399842e-05,
+      "loss": 0.2551,
+      "step": 1562
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.385777377498309e-05,
+      "loss": 0.2413,
+      "step": 1563
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.384974351357893e-05,
+      "loss": 0.2178,
+      "step": 1564
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.384170935971938e-05,
+      "loss": 0.2508,
+      "step": 1565
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3833671316938166e-05,
+      "loss": 0.2326,
+      "step": 1566
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.382562938877073e-05,
+      "loss": 0.2739,
+      "step": 1567
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3817583578754215e-05,
+      "loss": 0.2574,
+      "step": 1568
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3809533890427485e-05,
+      "loss": 0.2547,
+      "step": 1569
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3801480327331086e-05,
+      "loss": 0.219,
+      "step": 1570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3793422893007292e-05,
+      "loss": 0.2511,
+      "step": 1571
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.378536159100007e-05,
+      "loss": 0.3167,
+      "step": 1572
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.377729642485509e-05,
+      "loss": 0.2519,
+      "step": 1573
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3769227398119712e-05,
+      "loss": 0.2066,
+      "step": 1574
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.376115451434301e-05,
+      "loss": 0.2622,
+      "step": 1575
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3753077777075746e-05,
+      "loss": 0.2597,
+      "step": 1576
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3744997189870376e-05,
+      "loss": 0.2434,
+      "step": 1577
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3736912756281054e-05,
+      "loss": 0.2719,
+      "step": 1578
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.372882447986362e-05,
+      "loss": 0.2941,
+      "step": 1579
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3720732364175604e-05,
+      "loss": 0.2599,
+      "step": 1580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.371263641277623e-05,
+      "loss": 0.2682,
+      "step": 1581
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3704536629226416e-05,
+      "loss": 0.2414,
+      "step": 1582
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.369643301708874e-05,
+      "loss": 0.2868,
+      "step": 1583
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3688325579927494e-05,
+      "loss": 0.2493,
+      "step": 1584
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3680214321308636e-05,
+      "loss": 0.1997,
+      "step": 1585
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3672099244799803e-05,
+      "loss": 0.2325,
+      "step": 1586
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3663980353970316e-05,
+      "loss": 0.2199,
+      "step": 1587
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3655857652391173e-05,
+      "loss": 0.2575,
+      "step": 1588
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3647731143635055e-05,
+      "loss": 0.1657,
+      "step": 1589
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3639600831276305e-05,
+      "loss": 0.1801,
+      "step": 1590
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3631466718890946e-05,
+      "loss": 0.2179,
+      "step": 1591
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3623328810056674e-05,
+      "loss": 0.2074,
+      "step": 1592
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.361518710835285e-05,
+      "loss": 0.2676,
+      "step": 1593
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.360704161736051e-05,
+      "loss": 0.2473,
+      "step": 1594
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.359889234066234e-05,
+      "loss": 0.2146,
+      "step": 1595
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.359073928184272e-05,
+      "loss": 0.2449,
+      "step": 1596
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3582582444487665e-05,
+      "loss": 0.2423,
+      "step": 1597
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.357442183218487e-05,
+      "loss": 0.2358,
+      "step": 1598
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3566257448523675e-05,
+      "loss": 0.2138,
+      "step": 1599
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3558089297095103e-05,
+      "loss": 0.1892,
+      "step": 1600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3549917381491804e-05,
+      "loss": 0.2073,
+      "step": 1601
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.354174170530811e-05,
+      "loss": 0.2942,
+      "step": 1602
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3533562272139987e-05,
+      "loss": 0.2368,
+      "step": 1603
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.352537908558507e-05,
+      "loss": 0.269,
+      "step": 1604
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3517192149242632e-05,
+      "loss": 0.1936,
+      "step": 1605
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3509001466713603e-05,
+      "loss": 0.2351,
+      "step": 1606
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.350080704160056e-05,
+      "loss": 0.2803,
+      "step": 1607
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3492608877507717e-05,
+      "loss": 0.2436,
+      "step": 1608
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3484406978040946e-05,
+      "loss": 0.2349,
+      "step": 1609
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3476201346807754e-05,
+      "loss": 0.2103,
+      "step": 1610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.346799198741729e-05,
+      "loss": 0.2039,
+      "step": 1611
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.345977890348035e-05,
+      "loss": 0.1973,
+      "step": 1612
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3451562098609353e-05,
+      "loss": 0.2271,
+      "step": 1613
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.344334157641837e-05,
+      "loss": 0.2678,
+      "step": 1614
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.34351173405231e-05,
+      "loss": 0.2414,
+      "step": 1615
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3426889394540873e-05,
+      "loss": 0.2549,
+      "step": 1616
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3418657742090663e-05,
+      "loss": 0.1526,
+      "step": 1617
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3410422386793054e-05,
+      "loss": 0.1606,
+      "step": 1618
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3402183332270277e-05,
+      "loss": 0.2217,
+      "step": 1619
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3393940582146177e-05,
+      "loss": 0.2019,
+      "step": 1620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3385694140046238e-05,
+      "loss": 0.2135,
+      "step": 1621
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3377444009597555e-05,
+      "loss": 0.1898,
+      "step": 1622
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3369190194428857e-05,
+      "loss": 0.2139,
+      "step": 1623
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3360932698170477e-05,
+      "loss": 0.1651,
+      "step": 1624
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3352671524454386e-05,
+      "loss": 0.2255,
+      "step": 1625
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.334440667691416e-05,
+      "loss": 0.2132,
+      "step": 1626
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3336138159184998e-05,
+      "loss": 0.2263,
+      "step": 1627
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3327865974903706e-05,
+      "loss": 0.2683,
+      "step": 1628
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.331959012770871e-05,
+      "loss": 0.2681,
+      "step": 1629
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.331131062124004e-05,
+      "loss": 0.2009,
+      "step": 1630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3303027459139343e-05,
+      "loss": 0.2585,
+      "step": 1631
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.329474064504987e-05,
+      "loss": 0.2301,
+      "step": 1632
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.328645018261648e-05,
+      "loss": 0.2255,
+      "step": 1633
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.327815607548563e-05,
+      "loss": 0.2059,
+      "step": 1634
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3269858327305387e-05,
+      "loss": 0.2432,
+      "step": 1635
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.326155694172542e-05,
+      "loss": 0.2288,
+      "step": 1636
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3253251922396995e-05,
+      "loss": 0.2919,
+      "step": 1637
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3244943272972975e-05,
+      "loss": 0.2397,
+      "step": 1638
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3236630997107828e-05,
+      "loss": 0.2151,
+      "step": 1639
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.32283150984576e-05,
+      "loss": 0.2392,
+      "step": 1640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3219995580679945e-05,
+      "loss": 0.1807,
+      "step": 1641
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3211672447434113e-05,
+      "loss": 0.159,
+      "step": 1642
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3203345702380925e-05,
+      "loss": 0.2214,
+      "step": 1643
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3195015349182805e-05,
+      "loss": 0.2307,
+      "step": 1644
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3186681391503765e-05,
+      "loss": 0.2256,
+      "step": 1645
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3178343833009398e-05,
+      "loss": 0.188,
+      "step": 1646
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.317000267736687e-05,
+      "loss": 0.227,
+      "step": 1647
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3161657928244953e-05,
+      "loss": 0.2613,
+      "step": 1648
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.315330958931398e-05,
+      "loss": 0.2487,
+      "step": 1649
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.314495766424587e-05,
+      "loss": 0.2168,
+      "step": 1650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.313660215671412e-05,
+      "loss": 0.1985,
+      "step": 1651
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3128243070393802e-05,
+      "loss": 0.2353,
+      "step": 1652
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.311988040896156e-05,
+      "loss": 0.2006,
+      "step": 1653
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.311151417609561e-05,
+      "loss": 0.2773,
+      "step": 1654
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.310314437547575e-05,
+      "loss": 0.273,
+      "step": 1655
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3094771010783328e-05,
+      "loss": 0.2883,
+      "step": 1656
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3086394085701272e-05,
+      "loss": 0.2925,
+      "step": 1657
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.307801360391408e-05,
+      "loss": 0.2095,
+      "step": 1658
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3069629569107796e-05,
+      "loss": 0.2391,
+      "step": 1659
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3061241984970055e-05,
+      "loss": 0.2195,
+      "step": 1660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3052850855190024e-05,
+      "loss": 0.2102,
+      "step": 1661
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3044456183458448e-05,
+      "loss": 0.2693,
+      "step": 1662
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3036057973467628e-05,
+      "loss": 0.2011,
+      "step": 1663
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3027656228911406e-05,
+      "loss": 0.2474,
+      "step": 1664
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3019250953485204e-05,
+      "loss": 0.2739,
+      "step": 1665
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.301084215088597e-05,
+      "loss": 0.2008,
+      "step": 1666
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.300242982481223e-05,
+      "loss": 0.2708,
+      "step": 1667
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2994013978964034e-05,
+      "loss": 0.2226,
+      "step": 1668
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2985594617043003e-05,
+      "loss": 0.1999,
+      "step": 1669
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2977171742752287e-05,
+      "loss": 0.1966,
+      "step": 1670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2968745359796593e-05,
+      "loss": 0.2408,
+      "step": 1671
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2960315471882162e-05,
+      "loss": 0.1707,
+      "step": 1672
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2951882082716783e-05,
+      "loss": 0.247,
+      "step": 1673
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2943445196009785e-05,
+      "loss": 0.1886,
+      "step": 1674
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2935004815472026e-05,
+      "loss": 0.2696,
+      "step": 1675
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.292656094481592e-05,
+      "loss": 0.1881,
+      "step": 1676
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2918113587755395e-05,
+      "loss": 0.23,
+      "step": 1677
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2909662748005923e-05,
+      "loss": 0.1819,
+      "step": 1678
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2901208429284502e-05,
+      "loss": 0.1908,
+      "step": 1679
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2892750635309675e-05,
+      "loss": 0.2043,
+      "step": 1680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2884289369801497e-05,
+      "loss": 0.2103,
+      "step": 1681
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.287582463648155e-05,
+      "loss": 0.2349,
+      "step": 1682
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2867356439072956e-05,
+      "loss": 0.1871,
+      "step": 1683
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2858884781300353e-05,
+      "loss": 0.251,
+      "step": 1684
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2850409666889892e-05,
+      "loss": 0.2591,
+      "step": 1685
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2841931099569255e-05,
+      "loss": 0.2088,
+      "step": 1686
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2833449083067645e-05,
+      "loss": 0.2471,
+      "step": 1687
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.282496362111577e-05,
+      "loss": 0.2122,
+      "step": 1688
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2816474717445857e-05,
+      "loss": 0.1983,
+      "step": 1689
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2807982375791666e-05,
+      "loss": 0.1854,
+      "step": 1690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2799486599888445e-05,
+      "loss": 0.2668,
+      "step": 1691
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2790987393472954e-05,
+      "loss": 0.2189,
+      "step": 1692
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.278248476028348e-05,
+      "loss": 0.2132,
+      "step": 1693
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2773978704059792e-05,
+      "loss": 0.2528,
+      "step": 1694
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.27654692285432e-05,
+      "loss": 0.2101,
+      "step": 1695
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2756956337476468e-05,
+      "loss": 0.2187,
+      "step": 1696
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.274844003460392e-05,
+      "loss": 0.284,
+      "step": 1697
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2739920323671334e-05,
+      "loss": 0.2201,
+      "step": 1698
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2731397208426005e-05,
+      "loss": 0.2201,
+      "step": 1699
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2722870692616727e-05,
+      "loss": 0.2599,
+      "step": 1700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2714340779993793e-05,
+      "loss": 0.2213,
+      "step": 1701
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.270580747430898e-05,
+      "loss": 0.2362,
+      "step": 1702
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2697270779315565e-05,
+      "loss": 0.2413,
+      "step": 1703
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2688730698768304e-05,
+      "loss": 0.2401,
+      "step": 1704
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.268018723642346e-05,
+      "loss": 0.2709,
+      "step": 1705
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2671640396038773e-05,
+      "loss": 0.2922,
+      "step": 1706
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2663090181373463e-05,
+      "loss": 0.2514,
+      "step": 1707
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2654536596188253e-05,
+      "loss": 0.2223,
+      "step": 1708
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2645979644245327e-05,
+      "loss": 0.2572,
+      "step": 1709
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.263741932930836e-05,
+      "loss": 0.196,
+      "step": 1710
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2628855655142505e-05,
+      "loss": 0.2108,
+      "step": 1711
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2620288625514402e-05,
+      "loss": 0.2136,
+      "step": 1712
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.261171824419215e-05,
+      "loss": 0.2107,
+      "step": 1713
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2603144514945333e-05,
+      "loss": 0.2514,
+      "step": 1714
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2594567441545003e-05,
+      "loss": 0.2248,
+      "step": 1715
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2585987027763685e-05,
+      "loss": 0.1706,
+      "step": 1716
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.257740327737538e-05,
+      "loss": 0.2538,
+      "step": 1717
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2568816194155534e-05,
+      "loss": 0.2483,
+      "step": 1718
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.256022578188109e-05,
+      "loss": 0.1962,
+      "step": 1719
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.255163204433043e-05,
+      "loss": 0.1849,
+      "step": 1720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2543034985283406e-05,
+      "loss": 0.226,
+      "step": 1721
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.253443460852135e-05,
+      "loss": 0.2002,
+      "step": 1722
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2525830917827015e-05,
+      "loss": 0.2533,
+      "step": 1723
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2517223916984646e-05,
+      "loss": 0.2164,
+      "step": 1724
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2508613609779926e-05,
+      "loss": 0.2609,
+      "step": 1725
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.25e-05,
+      "loss": 0.2824,
+      "step": 1726
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2491383091433462e-05,
+      "loss": 0.1954,
+      "step": 1727
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2482762887870358e-05,
+      "loss": 0.221,
+      "step": 1728
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2474139393102182e-05,
+      "loss": 0.2161,
+      "step": 1729
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2465512610921876e-05,
+      "loss": 0.2079,
+      "step": 1730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2456882545123824e-05,
+      "loss": 0.2459,
+      "step": 1731
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2448249199503866e-05,
+      "loss": 0.1685,
+      "step": 1732
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2439612577859277e-05,
+      "loss": 0.1727,
+      "step": 1733
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2430972683988774e-05,
+      "loss": 0.216,
+      "step": 1734
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2422329521692508e-05,
+      "loss": 0.2254,
+      "step": 1735
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.241368309477207e-05,
+      "loss": 0.194,
+      "step": 1736
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.24050334070305e-05,
+      "loss": 0.241,
+      "step": 1737
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.239638046227225e-05,
+      "loss": 0.1963,
+      "step": 1738
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.238772426430322e-05,
+      "loss": 0.2144,
+      "step": 1739
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2379064816930737e-05,
+      "loss": 0.2094,
+      "step": 1740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2370402123963557e-05,
+      "loss": 0.2236,
+      "step": 1741
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.236173618921186e-05,
+      "loss": 0.2325,
+      "step": 1742
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.235306701648726e-05,
+      "loss": 0.2046,
+      "step": 1743
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.234439460960279e-05,
+      "loss": 0.2045,
+      "step": 1744
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.23357189723729e-05,
+      "loss": 0.2092,
+      "step": 1745
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2327040108613477e-05,
+      "loss": 0.2412,
+      "step": 1746
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2318358022141812e-05,
+      "loss": 0.1749,
+      "step": 1747
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.230967271677661e-05,
+      "loss": 0.2742,
+      "step": 1748
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2300984196338017e-05,
+      "loss": 0.1974,
+      "step": 1749
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.229229246464756e-05,
+      "loss": 0.2385,
+      "step": 1750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.228359752552821e-05,
+      "loss": 0.2481,
+      "step": 1751
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2274899382804316e-05,
+      "loss": 0.2636,
+      "step": 1752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2266198040301667e-05,
+      "loss": 0.2535,
+      "step": 1753
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2257493501847444e-05,
+      "loss": 0.2254,
+      "step": 1754
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2248785771270233e-05,
+      "loss": 0.1989,
+      "step": 1755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.224007485240003e-05,
+      "loss": 0.1907,
+      "step": 1756
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2231360749068223e-05,
+      "loss": 0.2099,
+      "step": 1757
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.222264346510761e-05,
+      "loss": 0.2299,
+      "step": 1758
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.221392300435239e-05,
+      "loss": 0.2234,
+      "step": 1759
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.220519937063815e-05,
+      "loss": 0.2631,
+      "step": 1760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.219647256780189e-05,
+      "loss": 0.2842,
+      "step": 1761
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.218774259968197e-05,
+      "loss": 0.1793,
+      "step": 1762
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.217900947011818e-05,
+      "loss": 0.1918,
+      "step": 1763
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2170273182951676e-05,
+      "loss": 0.2596,
+      "step": 1764
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2161533742025017e-05,
+      "loss": 0.2137,
+      "step": 1765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.215279115118214e-05,
+      "loss": 0.2011,
+      "step": 1766
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.214404541426837e-05,
+      "loss": 0.2101,
+      "step": 1767
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.213529653513042e-05,
+      "loss": 0.2332,
+      "step": 1768
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2126544517616375e-05,
+      "loss": 0.2093,
+      "step": 1769
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2117789365575714e-05,
+      "loss": 0.1978,
+      "step": 1770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2109031082859276e-05,
+      "loss": 0.1585,
+      "step": 1771
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2100269673319305e-05,
+      "loss": 0.1492,
+      "step": 1772
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.209150514080939e-05,
+      "loss": 0.2303,
+      "step": 1773
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2082737489184515e-05,
+      "loss": 0.2062,
+      "step": 1774
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2073966722301016e-05,
+      "loss": 0.1897,
+      "step": 1775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.206519284401663e-05,
+      "loss": 0.2569,
+      "step": 1776
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2056415858190437e-05,
+      "loss": 0.2103,
+      "step": 1777
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.204763576868288e-05,
+      "loss": 0.2067,
+      "step": 1778
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2038852579355793e-05,
+      "loss": 0.2144,
+      "step": 1779
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.203006629407235e-05,
+      "loss": 0.1811,
+      "step": 1780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2021276916697097e-05,
+      "loss": 0.1875,
+      "step": 1781
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2012484451095933e-05,
+      "loss": 0.223,
+      "step": 1782
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.200368890113613e-05,
+      "loss": 0.2591,
+      "step": 1783
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.19948902706863e-05,
+      "loss": 0.219,
+      "step": 1784
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.198608856361642e-05,
+      "loss": 0.2182,
+      "step": 1785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.197728378379782e-05,
+      "loss": 0.1808,
+      "step": 1786
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.196847593510317e-05,
+      "loss": 0.1756,
+      "step": 1787
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1959665021406508e-05,
+      "loss": 0.2461,
+      "step": 1788
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.19508510465832e-05,
+      "loss": 0.2896,
+      "step": 1789
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1942034014509977e-05,
+      "loss": 0.1628,
+      "step": 1790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1933213929064904e-05,
+      "loss": 0.173,
+      "step": 1791
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1924390794127397e-05,
+      "loss": 0.2608,
+      "step": 1792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1915564613578198e-05,
+      "loss": 0.2313,
+      "step": 1793
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1906735391299403e-05,
+      "loss": 0.2171,
+      "step": 1794
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1897903131174447e-05,
+      "loss": 0.2516,
+      "step": 1795
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1889067837088082e-05,
+      "loss": 0.3097,
+      "step": 1796
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1880229512926426e-05,
+      "loss": 0.1685,
+      "step": 1797
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1871388162576898e-05,
+      "loss": 0.2005,
+      "step": 1798
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1862543789928263e-05,
+      "loss": 0.1892,
+      "step": 1799
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1853696398870623e-05,
+      "loss": 0.2066,
+      "step": 1800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1844845993295386e-05,
+      "loss": 0.2408,
+      "step": 1801
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1835992577095314e-05,
+      "loss": 0.2805,
+      "step": 1802
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1827136154164457e-05,
+      "loss": 0.2025,
+      "step": 1803
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1818276728398233e-05,
+      "loss": 0.2491,
+      "step": 1804
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1809414303693336e-05,
+      "loss": 0.2155,
+      "step": 1805
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.180054888394781e-05,
+      "loss": 0.2438,
+      "step": 1806
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1791680473060998e-05,
+      "loss": 0.2305,
+      "step": 1807
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1782809074933568e-05,
+      "loss": 0.2534,
+      "step": 1808
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.177393469346751e-05,
+      "loss": 0.1507,
+      "step": 1809
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.17650573325661e-05,
+      "loss": 0.1728,
+      "step": 1810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1756176996133952e-05,
+      "loss": 0.2474,
+      "step": 1811
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1747293688076974e-05,
+      "loss": 0.1888,
+      "step": 1812
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1738407412302382e-05,
+      "loss": 0.1908,
+      "step": 1813
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1729518172718704e-05,
+      "loss": 0.1879,
+      "step": 1814
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1720625973235762e-05,
+      "loss": 0.2051,
+      "step": 1815
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1711730817764692e-05,
+      "loss": 0.2317,
+      "step": 1816
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1702832710217915e-05,
+      "loss": 0.2129,
+      "step": 1817
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1693931654509166e-05,
+      "loss": 0.2129,
+      "step": 1818
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.168502765455347e-05,
+      "loss": 0.1893,
+      "step": 1819
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.167612071426714e-05,
+      "loss": 0.1995,
+      "step": 1820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1667210837567787e-05,
+      "loss": 0.1488,
+      "step": 1821
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1658298028374323e-05,
+      "loss": 0.2651,
+      "step": 1822
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1649382290606942e-05,
+      "loss": 0.2392,
+      "step": 1823
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1640463628187124e-05,
+      "loss": 0.2112,
+      "step": 1824
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1631542045037632e-05,
+      "loss": 0.2471,
+      "step": 1825
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.162261754508253e-05,
+      "loss": 0.2242,
+      "step": 1826
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.161369013224714e-05,
+      "loss": 0.2007,
+      "step": 1827
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1604759810458093e-05,
+      "loss": 0.1971,
+      "step": 1828
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1595826583643284e-05,
+      "loss": 0.2735,
+      "step": 1829
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.158689045573188e-05,
+      "loss": 0.1758,
+      "step": 1830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1577951430654334e-05,
+      "loss": 0.2025,
+      "step": 1831
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.156900951234237e-05,
+      "loss": 0.1877,
+      "step": 1832
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1560064704728995e-05,
+      "loss": 0.2282,
+      "step": 1833
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1551117011748466e-05,
+      "loss": 0.2383,
+      "step": 1834
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1542166437336323e-05,
+      "loss": 0.2551,
+      "step": 1835
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1533212985429374e-05,
+      "loss": 0.1967,
+      "step": 1836
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.152425665996569e-05,
+      "loss": 0.2364,
+      "step": 1837
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.15152974648846e-05,
+      "loss": 0.2603,
+      "step": 1838
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.15063354041267e-05,
+      "loss": 0.2381,
+      "step": 1839
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.149737048163385e-05,
+      "loss": 0.178,
+      "step": 1840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.148840270134917e-05,
+      "loss": 0.1868,
+      "step": 1841
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1479432067217022e-05,
+      "loss": 0.2054,
+      "step": 1842
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.147045858318304e-05,
+      "loss": 0.2182,
+      "step": 1843
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1461482253194108e-05,
+      "loss": 0.1973,
+      "step": 1844
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1452503081198354e-05,
+      "loss": 0.269,
+      "step": 1845
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.144352107114516e-05,
+      "loss": 0.1994,
+      "step": 1846
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.143453622698516e-05,
+      "loss": 0.1527,
+      "step": 1847
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1425548552670233e-05,
+      "loss": 0.1935,
+      "step": 1848
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.14165580521535e-05,
+      "loss": 0.2373,
+      "step": 1849
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1407564729389327e-05,
+      "loss": 0.2106,
+      "step": 1850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.139856858833332e-05,
+      "loss": 0.2422,
+      "step": 1851
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1389569632942325e-05,
+      "loss": 0.1507,
+      "step": 1852
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1380567867174428e-05,
+      "loss": 0.2159,
+      "step": 1853
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.137156329498895e-05,
+      "loss": 0.2155,
+      "step": 1854
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.136255592034645e-05,
+      "loss": 0.186,
+      "step": 1855
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1353545747208713e-05,
+      "loss": 0.1926,
+      "step": 1856
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1344532779538746e-05,
+      "loss": 0.1924,
+      "step": 1857
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.133551702130081e-05,
+      "loss": 0.1829,
+      "step": 1858
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1326498476460377e-05,
+      "loss": 0.1898,
+      "step": 1859
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1317477148984148e-05,
+      "loss": 0.2291,
+      "step": 1860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1308453042840046e-05,
+      "loss": 0.2379,
+      "step": 1861
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.129942616199722e-05,
+      "loss": 0.2469,
+      "step": 1862
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.129039651042603e-05,
+      "loss": 0.2606,
+      "step": 1863
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1281364092098066e-05,
+      "loss": 0.1631,
+      "step": 1864
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1272328910986137e-05,
+      "loss": 0.2177,
+      "step": 1865
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1263290971064256e-05,
+      "loss": 0.1939,
+      "step": 1866
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.125425027630765e-05,
+      "loss": 0.1972,
+      "step": 1867
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.124520683069277e-05,
+      "loss": 0.3251,
+      "step": 1868
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1236160638197264e-05,
+      "loss": 0.1661,
+      "step": 1869
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1227111702799996e-05,
+      "loss": 0.1808,
+      "step": 1870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.121806002848103e-05,
+      "loss": 0.1962,
+      "step": 1871
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.120900561922164e-05,
+      "loss": 0.1932,
+      "step": 1872
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1199948479004306e-05,
+      "loss": 0.2449,
+      "step": 1873
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1190888611812692e-05,
+      "loss": 0.1855,
+      "step": 1874
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1181826021631685e-05,
+      "loss": 0.2145,
+      "step": 1875
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.117276071244736e-05,
+      "loss": 0.2126,
+      "step": 1876
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1163692688246975e-05,
+      "loss": 0.1789,
+      "step": 1877
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1154621953018998e-05,
+      "loss": 0.1916,
+      "step": 1878
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1145548510753096e-05,
+      "loss": 0.1923,
+      "step": 1879
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.11364723654401e-05,
+      "loss": 0.2189,
+      "step": 1880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1127393521072055e-05,
+      "loss": 0.2149,
+      "step": 1881
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1118311981642177e-05,
+      "loss": 0.2215,
+      "step": 1882
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1109227751144882e-05,
+      "loss": 0.1936,
+      "step": 1883
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1100140833575756e-05,
+      "loss": 0.144,
+      "step": 1884
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1091051232931575e-05,
+      "loss": 0.2362,
+      "step": 1885
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1081958953210297e-05,
+      "loss": 0.2863,
+      "step": 1886
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1072863998411045e-05,
+      "loss": 0.1752,
+      "step": 1887
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1063766372534133e-05,
+      "loss": 0.2302,
+      "step": 1888
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1054666079581036e-05,
+      "loss": 0.2467,
+      "step": 1889
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.104556312355443e-05,
+      "loss": 0.2242,
+      "step": 1890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1036457508458127e-05,
+      "loss": 0.2011,
+      "step": 1891
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1027349238297126e-05,
+      "loss": 0.1931,
+      "step": 1892
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1018238317077594e-05,
+      "loss": 0.216,
+      "step": 1893
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1009124748806867e-05,
+      "loss": 0.2296,
+      "step": 1894
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1000008537493434e-05,
+      "loss": 0.2199,
+      "step": 1895
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0990889687146953e-05,
+      "loss": 0.2034,
+      "step": 1896
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.098176820177825e-05,
+      "loss": 0.2485,
+      "step": 1897
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.09726440853993e-05,
+      "loss": 0.1728,
+      "step": 1898
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.096351734202323e-05,
+      "loss": 0.1725,
+      "step": 1899
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0954387975664345e-05,
+      "loss": 0.299,
+      "step": 1900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0945255990338075e-05,
+      "loss": 0.1915,
+      "step": 1901
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.093612139006103e-05,
+      "loss": 0.1748,
+      "step": 1902
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.092698417885094e-05,
+      "loss": 0.2037,
+      "step": 1903
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0917844360726713e-05,
+      "loss": 0.1786,
+      "step": 1904
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0908701939708385e-05,
+      "loss": 0.192,
+      "step": 1905
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.089955691981714e-05,
+      "loss": 0.2364,
+      "step": 1906
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0890409305075315e-05,
+      "loss": 0.2297,
+      "step": 1907
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0881259099506372e-05,
+      "loss": 0.183,
+      "step": 1908
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.087210630713493e-05,
+      "loss": 0.2309,
+      "step": 1909
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0862950931986723e-05,
+      "loss": 0.1895,
+      "step": 1910
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0853792978088648e-05,
+      "loss": 0.1919,
+      "step": 1911
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.084463244946872e-05,
+      "loss": 0.169,
+      "step": 1912
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.083546935015608e-05,
+      "loss": 0.223,
+      "step": 1913
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0826303684181023e-05,
+      "loss": 0.2206,
+      "step": 1914
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0817135455574956e-05,
+      "loss": 0.1941,
+      "step": 1915
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.080796466837041e-05,
+      "loss": 0.1533,
+      "step": 1916
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.079879132660105e-05,
+      "loss": 0.1922,
+      "step": 1917
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.078961543430167e-05,
+      "loss": 0.2367,
+      "step": 1918
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.078043699550817e-05,
+      "loss": 0.2149,
+      "step": 1919
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.077125601425758e-05,
+      "loss": 0.222,
+      "step": 1920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.076207249458804e-05,
+      "loss": 0.1935,
+      "step": 1921
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0752886440538825e-05,
+      "loss": 0.2066,
+      "step": 1922
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0743697856150306e-05,
+      "loss": 0.1642,
+      "step": 1923
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0734506745463972e-05,
+      "loss": 0.2298,
+      "step": 1924
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.072531311252243e-05,
+      "loss": 0.2277,
+      "step": 1925
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0716116961369388e-05,
+      "loss": 0.2147,
+      "step": 1926
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0706918296049666e-05,
+      "loss": 0.2162,
+      "step": 1927
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.069771712060918e-05,
+      "loss": 0.1327,
+      "step": 1928
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0688513439094974e-05,
+      "loss": 0.1932,
+      "step": 1929
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.067930725555517e-05,
+      "loss": 0.2465,
+      "step": 1930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0670098574038994e-05,
+      "loss": 0.1967,
+      "step": 1931
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.066088739859679e-05,
+      "loss": 0.2229,
+      "step": 1932
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0651673733279974e-05,
+      "loss": 0.2093,
+      "step": 1933
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0642457582141072e-05,
+      "loss": 0.2158,
+      "step": 1934
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0633238949233695e-05,
+      "loss": 0.2008,
+      "step": 1935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.062401783861256e-05,
+      "loss": 0.1623,
+      "step": 1936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0614794254333455e-05,
+      "loss": 0.2219,
+      "step": 1937
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0605568200453265e-05,
+      "loss": 0.195,
+      "step": 1938
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0596339681029968e-05,
+      "loss": 0.1975,
+      "step": 1939
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0587108700122614e-05,
+      "loss": 0.2319,
+      "step": 1940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.057787526179134e-05,
+      "loss": 0.1689,
+      "step": 1941
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0568639370097367e-05,
+      "loss": 0.1939,
+      "step": 1942
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0559401029102993e-05,
+      "loss": 0.135,
+      "step": 1943
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0550160242871592e-05,
+      "loss": 0.2117,
+      "step": 1944
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0540917015467623e-05,
+      "loss": 0.2135,
+      "step": 1945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.05316713509566e-05,
+      "loss": 0.132,
+      "step": 1946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.052242325340513e-05,
+      "loss": 0.2233,
+      "step": 1947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.051317272688087e-05,
+      "loss": 0.1817,
+      "step": 1948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0503919775452555e-05,
+      "loss": 0.2198,
+      "step": 1949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0494664403189994e-05,
+      "loss": 0.2271,
+      "step": 1950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.048540661416405e-05,
+      "loss": 0.2595,
+      "step": 1951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0476146412446652e-05,
+      "loss": 0.2301,
+      "step": 1952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0466883802110788e-05,
+      "loss": 0.2377,
+      "step": 1953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0457618787230515e-05,
+      "loss": 0.2289,
+      "step": 1954
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0448351371880933e-05,
+      "loss": 0.2252,
+      "step": 1955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.043908156013821e-05,
+      "loss": 0.2246,
+      "step": 1956
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0429809356079566e-05,
+      "loss": 0.1927,
+      "step": 1957
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.042053476378326e-05,
+      "loss": 0.2059,
+      "step": 1958
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.041125778732863e-05,
+      "loss": 0.1995,
+      "step": 1959
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.040197843079602e-05,
+      "loss": 0.227,
+      "step": 1960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0392696698266868e-05,
+      "loss": 0.2082,
+      "step": 1961
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0383412593823623e-05,
+      "loss": 0.2819,
+      "step": 1962
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0374126121549794e-05,
+      "loss": 0.2089,
+      "step": 1963
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.036483728552992e-05,
+      "loss": 0.2313,
+      "step": 1964
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0355546089849597e-05,
+      "loss": 0.2017,
+      "step": 1965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.034625253859544e-05,
+      "loss": 0.167,
+      "step": 1966
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0336956635855102e-05,
+      "loss": 0.2338,
+      "step": 1967
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0327658385717293e-05,
+      "loss": 0.2218,
+      "step": 1968
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.031835779227173e-05,
+      "loss": 0.1672,
+      "step": 1969
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.030905485960916e-05,
+      "loss": 0.2706,
+      "step": 1970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0299749591821384e-05,
+      "loss": 0.1846,
+      "step": 1971
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0290441993001207e-05,
+      "loss": 0.2079,
+      "step": 1972
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.028113206724246e-05,
+      "loss": 0.2795,
+      "step": 1973
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.027181981864002e-05,
+      "loss": 0.1775,
+      "step": 1974
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0262505251289757e-05,
+      "loss": 0.2472,
+      "step": 1975
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.025318836928858e-05,
+      "loss": 0.1681,
+      "step": 1976
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0243869176734407e-05,
+      "loss": 0.1688,
+      "step": 1977
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0234547677726175e-05,
+      "loss": 0.238,
+      "step": 1978
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0225223876363832e-05,
+      "loss": 0.1697,
+      "step": 1979
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.021589777674835e-05,
+      "loss": 0.1258,
+      "step": 1980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0206569382981704e-05,
+      "loss": 0.1108,
+      "step": 1981
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0197238699166867e-05,
+      "loss": 0.1934,
+      "step": 1982
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0187905729407844e-05,
+      "loss": 0.1994,
+      "step": 1983
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.017857047780962e-05,
+      "loss": 0.1901,
+      "step": 1984
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0169232948478205e-05,
+      "loss": 0.2044,
+      "step": 1985
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0159893145520593e-05,
+      "loss": 0.1933,
+      "step": 1986
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0150551073044797e-05,
+      "loss": 0.2247,
+      "step": 1987
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0141206735159803e-05,
+      "loss": 0.2277,
+      "step": 1988
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.013186013597562e-05,
+      "loss": 0.1917,
+      "step": 1989
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0122511279603234e-05,
+      "loss": 0.2139,
+      "step": 1990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.011316017015463e-05,
+      "loss": 0.2082,
+      "step": 1991
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0103806811742782e-05,
+      "loss": 0.2023,
+      "step": 1992
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.009445120848166e-05,
+      "loss": 0.2112,
+      "step": 1993
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0085093364486208e-05,
+      "loss": 0.1691,
+      "step": 1994
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.007573328387237e-05,
+      "loss": 0.206,
+      "step": 1995
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.006637097075706e-05,
+      "loss": 0.1734,
+      "step": 1996
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0057006429258196e-05,
+      "loss": 0.1778,
+      "step": 1997
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0047639663494644e-05,
+      "loss": 0.22,
+      "step": 1998
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.003827067758627e-05,
+      "loss": 0.1486,
+      "step": 1999
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0028899475653918e-05,
+      "loss": 0.1589,
+      "step": 2000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0019526061819397e-05,
+      "loss": 0.2651,
+      "step": 2001
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.001015044020549e-05,
+      "loss": 0.2541,
+      "step": 2002
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.000077261493596e-05,
+      "loss": 0.2249,
+      "step": 2003
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9991392590135523e-05,
+      "loss": 0.2639,
+      "step": 2004
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9982010369929886e-05,
+      "loss": 0.2662,
+      "step": 2005
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.99726259584457e-05,
+      "loss": 0.173,
+      "step": 2006
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9963239359810586e-05,
+      "loss": 0.2338,
+      "step": 2007
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.995385057815313e-05,
+      "loss": 0.2315,
+      "step": 2008
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9944459617602882e-05,
+      "loss": 0.1764,
+      "step": 2009
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.993506648229034e-05,
+      "loss": 0.1692,
+      "step": 2010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9925671176346966e-05,
+      "loss": 0.1626,
+      "step": 2011
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9916273703905173e-05,
+      "loss": 0.2151,
+      "step": 2012
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9906874069098332e-05,
+      "loss": 0.1942,
+      "step": 2013
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9897472276060754e-05,
+      "loss": 0.208,
+      "step": 2014
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9888068328927718e-05,
+      "loss": 0.1389,
+      "step": 2015
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.987866223183543e-05,
+      "loss": 0.1842,
+      "step": 2016
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9869253988921053e-05,
+      "loss": 0.1639,
+      "step": 2017
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9859843604322697e-05,
+      "loss": 0.2051,
+      "step": 2018
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.98504310821794e-05,
+      "loss": 0.2534,
+      "step": 2019
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9841016426631155e-05,
+      "loss": 0.1705,
+      "step": 2020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.983159964181889e-05,
+      "loss": 0.1935,
+      "step": 2021
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.982218073188446e-05,
+      "loss": 0.2665,
+      "step": 2022
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.981275970097067e-05,
+      "loss": 0.158,
+      "step": 2023
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9803336553221232e-05,
+      "loss": 0.2296,
+      "step": 2024
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9793911292780828e-05,
+      "loss": 0.1898,
+      "step": 2025
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9784483923795037e-05,
+      "loss": 0.1749,
+      "step": 2026
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.977505445041037e-05,
+      "loss": 0.1302,
+      "step": 2027
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9765622876774283e-05,
+      "loss": 0.1939,
+      "step": 2028
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.975618920703514e-05,
+      "loss": 0.234,
+      "step": 2029
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.974675344534222e-05,
+      "loss": 0.2279,
+      "step": 2030
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.973731559584574e-05,
+      "loss": 0.179,
+      "step": 2031
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9727875662696822e-05,
+      "loss": 0.1717,
+      "step": 2032
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.971843365004751e-05,
+      "loss": 0.1715,
+      "step": 2033
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.970898956205077e-05,
+      "loss": 0.1992,
+      "step": 2034
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9699543402860458e-05,
+      "loss": 0.1969,
+      "step": 2035
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9690095176631373e-05,
+      "loss": 0.1795,
+      "step": 2036
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9680644887519194e-05,
+      "loss": 0.1643,
+      "step": 2037
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9671192539680524e-05,
+      "loss": 0.1783,
+      "step": 2038
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9661738137272862e-05,
+      "loss": 0.2446,
+      "step": 2039
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9652281684454622e-05,
+      "loss": 0.2157,
+      "step": 2040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9642823185385108e-05,
+      "loss": 0.2095,
+      "step": 2041
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9633362644224532e-05,
+      "loss": 0.1615,
+      "step": 2042
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9623900065134e-05,
+      "loss": 0.2083,
+      "step": 2043
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.961443545227552e-05,
+      "loss": 0.2434,
+      "step": 2044
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9604968809811987e-05,
+      "loss": 0.1196,
+      "step": 2045
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.959550014190719e-05,
+      "loss": 0.2002,
+      "step": 2046
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9586029452725813e-05,
+      "loss": 0.1988,
+      "step": 2047
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.957655674643343e-05,
+      "loss": 0.1948,
+      "step": 2048
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9567082027196495e-05,
+      "loss": 0.1838,
+      "step": 2049
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9557605299182355e-05,
+      "loss": 0.1548,
+      "step": 2050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.954812656655924e-05,
+      "loss": 0.1893,
+      "step": 2051
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.953864583349625e-05,
+      "loss": 0.1856,
+      "step": 2052
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9529163104163373e-05,
+      "loss": 0.1945,
+      "step": 2053
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9519678382731483e-05,
+      "loss": 0.1534,
+      "step": 2054
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.951019167337233e-05,
+      "loss": 0.206,
+      "step": 2055
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9500702980258507e-05,
+      "loss": 0.1976,
+      "step": 2056
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9491212307563528e-05,
+      "loss": 0.2431,
+      "step": 2057
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9481719659461742e-05,
+      "loss": 0.1558,
+      "step": 2058
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9472225040128384e-05,
+      "loss": 0.2015,
+      "step": 2059
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9462728453739543e-05,
+      "loss": 0.2092,
+      "step": 2060
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.945322990447219e-05,
+      "loss": 0.214,
+      "step": 2061
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.944372939650414e-05,
+      "loss": 0.2391,
+      "step": 2062
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9434226934014085e-05,
+      "loss": 0.24,
+      "step": 2063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9424722521181572e-05,
+      "loss": 0.1745,
+      "step": 2064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9415216162187e-05,
+      "loss": 0.1805,
+      "step": 2065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9405707861211634e-05,
+      "loss": 0.1629,
+      "step": 2066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9396197622437588e-05,
+      "loss": 0.1358,
+      "step": 2067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9386685450047827e-05,
+      "loss": 0.184,
+      "step": 2068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9377171348226167e-05,
+      "loss": 0.1378,
+      "step": 2069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9367655321157276e-05,
+      "loss": 0.1934,
+      "step": 2070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.935813737302666e-05,
+      "loss": 0.2073,
+      "step": 2071
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9348617508020685e-05,
+      "loss": 0.188,
+      "step": 2072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9339095730326548e-05,
+      "loss": 0.1998,
+      "step": 2073
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9329572044132286e-05,
+      "loss": 0.2119,
+      "step": 2074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.932004645362679e-05,
+      "loss": 0.2493,
+      "step": 2075
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9310518962999774e-05,
+      "loss": 0.2288,
+      "step": 2076
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9300989576441788e-05,
+      "loss": 0.2037,
+      "step": 2077
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.929145829814422e-05,
+      "loss": 0.182,
+      "step": 2078
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9281925132299297e-05,
+      "loss": 0.1777,
+      "step": 2079
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.927239008310007e-05,
+      "loss": 0.2272,
+      "step": 2080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9262853154740413e-05,
+      "loss": 0.1914,
+      "step": 2081
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9253314351415033e-05,
+      "loss": 0.2039,
+      "step": 2082
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9243773677319464e-05,
+      "loss": 0.1609,
+      "step": 2083
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9234231136650054e-05,
+      "loss": 0.191,
+      "step": 2084
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9224686733603974e-05,
+      "loss": 0.1558,
+      "step": 2085
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9215140472379227e-05,
+      "loss": 0.226,
+      "step": 2086
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.920559235717462e-05,
+      "loss": 0.233,
+      "step": 2087
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9196042392189772e-05,
+      "loss": 0.1878,
+      "step": 2088
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9186490581625134e-05,
+      "loss": 0.1505,
+      "step": 2089
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9176936929681948e-05,
+      "loss": 0.245,
+      "step": 2090
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.916738144056228e-05,
+      "loss": 0.2067,
+      "step": 2091
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9157824118468996e-05,
+      "loss": 0.1593,
+      "step": 2092
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9148264967605776e-05,
+      "loss": 0.1946,
+      "step": 2093
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9138703992177093e-05,
+      "loss": 0.2437,
+      "step": 2094
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.912914119638824e-05,
+      "loss": 0.2455,
+      "step": 2095
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.911957658444529e-05,
+      "loss": 0.2091,
+      "step": 2096
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9110010160555128e-05,
+      "loss": 0.2461,
+      "step": 2097
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9100441928925437e-05,
+      "loss": 0.2313,
+      "step": 2098
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.909087189376468e-05,
+      "loss": 0.2217,
+      "step": 2099
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9081300059282135e-05,
+      "loss": 0.1633,
+      "step": 2100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.907172642968786e-05,
+      "loss": 0.1878,
+      "step": 2101
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.90621510091927e-05,
+      "loss": 0.1998,
+      "step": 2102
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9052573802008285e-05,
+      "loss": 0.1896,
+      "step": 2103
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9042994812347047e-05,
+      "loss": 0.1546,
+      "step": 2104
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.903341404442219e-05,
+      "loss": 0.2287,
+      "step": 2105
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9023831502447693e-05,
+      "loss": 0.1901,
+      "step": 2106
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9014247190638338e-05,
+      "loss": 0.1491,
+      "step": 2107
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9004661113209658e-05,
+      "loss": 0.1988,
+      "step": 2108
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8995073274377987e-05,
+      "loss": 0.1557,
+      "step": 2109
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.898548367836041e-05,
+      "loss": 0.2014,
+      "step": 2110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8975892329374813e-05,
+      "loss": 0.1469,
+      "step": 2111
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896629923163983e-05,
+      "loss": 0.1846,
+      "step": 2112
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8956704389374868e-05,
+      "loss": 0.197,
+      "step": 2113
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8947107806800115e-05,
+      "loss": 0.2136,
+      "step": 2114
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.893750948813651e-05,
+      "loss": 0.2268,
+      "step": 2115
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892790943760576e-05,
+      "loss": 0.2304,
+      "step": 2116
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8918307659430332e-05,
+      "loss": 0.1724,
+      "step": 2117
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8908704157833458e-05,
+      "loss": 0.2363,
+      "step": 2118
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8899098937039127e-05,
+      "loss": 0.2224,
+      "step": 2119
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.888949200127207e-05,
+      "loss": 0.1493,
+      "step": 2120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8879883354757804e-05,
+      "loss": 0.2191,
+      "step": 2121
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8870273001722564e-05,
+      "loss": 0.2135,
+      "step": 2122
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8860660946393362e-05,
+      "loss": 0.2062,
+      "step": 2123
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.885104719299793e-05,
+      "loss": 0.1911,
+      "step": 2124
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.884143174576478e-05,
+      "loss": 0.1757,
+      "step": 2125
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.883181460892315e-05,
+      "loss": 0.172,
+      "step": 2126
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.882219578670302e-05,
+      "loss": 0.1685,
+      "step": 2127
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.881257528333511e-05,
+      "loss": 0.1661,
+      "step": 2128
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8802953103050903e-05,
+      "loss": 0.2069,
+      "step": 2129
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8793329250082593e-05,
+      "loss": 0.1571,
+      "step": 2130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.878370372866311e-05,
+      "loss": 0.1687,
+      "step": 2131
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8774076543026136e-05,
+      "loss": 0.1465,
+      "step": 2132
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8764447697406073e-05,
+      "loss": 0.2028,
+      "step": 2133
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8754817196038053e-05,
+      "loss": 0.1732,
+      "step": 2134
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.874518504315794e-05,
+      "loss": 0.2481,
+      "step": 2135
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.873555124300232e-05,
+      "loss": 0.2239,
+      "step": 2136
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.872591579980851e-05,
+      "loss": 0.2115,
+      "step": 2137
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8716278717814546e-05,
+      "loss": 0.1931,
+      "step": 2138
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8706640001259186e-05,
+      "loss": 0.1877,
+      "step": 2139
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8696999654381905e-05,
+      "loss": 0.1441,
+      "step": 2140
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8687357681422894e-05,
+      "loss": 0.1596,
+      "step": 2141
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.867771408662306e-05,
+      "loss": 0.1714,
+      "step": 2142
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8668068874224027e-05,
+      "loss": 0.1934,
+      "step": 2143
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.865842204846813e-05,
+      "loss": 0.15,
+      "step": 2144
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8648773613598405e-05,
+      "loss": 0.1459,
+      "step": 2145
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8639123573858613e-05,
+      "loss": 0.2533,
+      "step": 2146
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.86294719334932e-05,
+      "loss": 0.1573,
+      "step": 2147
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8619818696747337e-05,
+      "loss": 0.1706,
+      "step": 2148
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.861016386786687e-05,
+      "loss": 0.2371,
+      "step": 2149
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.860050745109838e-05,
+      "loss": 0.2471,
+      "step": 2150
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8590849450689113e-05,
+      "loss": 0.18,
+      "step": 2151
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8581189870887036e-05,
+      "loss": 0.2098,
+      "step": 2152
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.85715287159408e-05,
+      "loss": 0.2076,
+      "step": 2153
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.856186599009974e-05,
+      "loss": 0.2033,
+      "step": 2154
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.855220169761391e-05,
+      "loss": 0.1585,
+      "step": 2155
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.854253584273402e-05,
+      "loss": 0.2826,
+      "step": 2156
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8532868429711492e-05,
+      "loss": 0.2123,
+      "step": 2157
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8523199462798422e-05,
+      "loss": 0.243,
+      "step": 2158
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.851352894624759e-05,
+      "loss": 0.1299,
+      "step": 2159
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8503856884312453e-05,
+      "loss": 0.1814,
+      "step": 2160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.849418328124716e-05,
+      "loss": 0.1776,
+      "step": 2161
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8484508141306533e-05,
+      "loss": 0.2142,
+      "step": 2162
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8474831468746057e-05,
+      "loss": 0.1668,
+      "step": 2163
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.846515326782192e-05,
+      "loss": 0.1869,
+      "step": 2164
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.845547354279095e-05,
+      "loss": 0.1288,
+      "step": 2165
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8445792297910666e-05,
+      "loss": 0.186,
+      "step": 2166
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8436109537439246e-05,
+      "loss": 0.2778,
+      "step": 2167
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.842642526563554e-05,
+      "loss": 0.2069,
+      "step": 2168
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8416739486759067e-05,
+      "loss": 0.2123,
+      "step": 2169
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.840705220506999e-05,
+      "loss": 0.1691,
+      "step": 2170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8397363424829158e-05,
+      "loss": 0.1591,
+      "step": 2171
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.838767315029806e-05,
+      "loss": 0.2084,
+      "step": 2172
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8377981385738852e-05,
+      "loss": 0.2132,
+      "step": 2173
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8368288135414336e-05,
+      "loss": 0.1978,
+      "step": 2174
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8358593403587982e-05,
+      "loss": 0.2131,
+      "step": 2175
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8348897194523903e-05,
+      "loss": 0.1443,
+      "step": 2176
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8339199512486856e-05,
+      "loss": 0.1356,
+      "step": 2177
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8329500361742257e-05,
+      "loss": 0.1719,
+      "step": 2178
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8319799746556162e-05,
+      "loss": 0.2167,
+      "step": 2179
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.831009767119527e-05,
+      "loss": 0.1763,
+      "step": 2180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8300394139926928e-05,
+      "loss": 0.2912,
+      "step": 2181
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.829068915701912e-05,
+      "loss": 0.1788,
+      "step": 2182
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.828098272674047e-05,
+      "loss": 0.23,
+      "step": 2183
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8271274853360224e-05,
+      "loss": 0.1613,
+      "step": 2184
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8261565541148292e-05,
+      "loss": 0.2105,
+      "step": 2185
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8251854794375194e-05,
+      "loss": 0.1921,
+      "step": 2186
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.824214261731209e-05,
+      "loss": 0.224,
+      "step": 2187
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.823242901423076e-05,
+      "loss": 0.1952,
+      "step": 2188
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8222713989403624e-05,
+      "loss": 0.2131,
+      "step": 2189
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8212997547103722e-05,
+      "loss": 0.1559,
+      "step": 2190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8203279691604714e-05,
+      "loss": 0.1625,
+      "step": 2191
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8193560427180882e-05,
+      "loss": 0.192,
+      "step": 2192
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8183839758107133e-05,
+      "loss": 0.2244,
+      "step": 2193
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8174117688658988e-05,
+      "loss": 0.1535,
+      "step": 2194
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8164394223112583e-05,
+      "loss": 0.1561,
+      "step": 2195
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8154669365744672e-05,
+      "loss": 0.2067,
+      "step": 2196
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.814494312083262e-05,
+      "loss": 0.2134,
+      "step": 2197
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.81352154926544e-05,
+      "loss": 0.1898,
+      "step": 2198
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.812548648548859e-05,
+      "loss": 0.1931,
+      "step": 2199
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8115756103614383e-05,
+      "loss": 0.1738,
+      "step": 2200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.810602435131158e-05,
+      "loss": 0.1582,
+      "step": 2201
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.809629123286056e-05,
+      "loss": 0.1989,
+      "step": 2202
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.808655675254234e-05,
+      "loss": 0.2064,
+      "step": 2203
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8076820914638503e-05,
+      "loss": 0.2051,
+      "step": 2204
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.806708372343125e-05,
+      "loss": 0.1478,
+      "step": 2205
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8057345183203364e-05,
+      "loss": 0.185,
+      "step": 2206
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.804760529823823e-05,
+      "loss": 0.1442,
+      "step": 2207
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8037864072819823e-05,
+      "loss": 0.2015,
+      "step": 2208
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8028121511232704e-05,
+      "loss": 0.2006,
+      "step": 2209
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8018377617762024e-05,
+      "loss": 0.2147,
+      "step": 2210
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8008632396693522e-05,
+      "loss": 0.144,
+      "step": 2211
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7998885852313517e-05,
+      "loss": 0.1619,
+      "step": 2212
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.798913798890891e-05,
+      "loss": 0.1724,
+      "step": 2213
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7979388810767184e-05,
+      "loss": 0.1309,
+      "step": 2214
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7969638322176404e-05,
+      "loss": 0.1916,
+      "step": 2215
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.795988652742521e-05,
+      "loss": 0.2238,
+      "step": 2216
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.79501334308028e-05,
+      "loss": 0.2,
+      "step": 2217
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.794037903659897e-05,
+      "loss": 0.1472,
+      "step": 2218
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7930623349104077e-05,
+      "loss": 0.2221,
+      "step": 2219
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7920866372609038e-05,
+      "loss": 0.1334,
+      "step": 2220
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.791110811140535e-05,
+      "loss": 0.2103,
+      "step": 2221
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7901348569785066e-05,
+      "loss": 0.1632,
+      "step": 2222
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7891587752040805e-05,
+      "loss": 0.1508,
+      "step": 2223
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7881825662465746e-05,
+      "loss": 0.2375,
+      "step": 2224
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7872062305353637e-05,
+      "loss": 0.2282,
+      "step": 2225
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.786229768499877e-05,
+      "loss": 0.2242,
+      "step": 2226
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.785253180569599e-05,
+      "loss": 0.181,
+      "step": 2227
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7842764671740722e-05,
+      "loss": 0.2089,
+      "step": 2228
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7832996287428914e-05,
+      "loss": 0.1847,
+      "step": 2229
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.782322665705707e-05,
+      "loss": 0.2155,
+      "step": 2230
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.781345578492226e-05,
+      "loss": 0.1924,
+      "step": 2231
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7803683675322073e-05,
+      "loss": 0.1807,
+      "step": 2232
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7793910332554666e-05,
+      "loss": 0.2022,
+      "step": 2233
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7784135760918723e-05,
+      "loss": 0.2575,
+      "step": 2234
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.777435996471348e-05,
+      "loss": 0.159,
+      "step": 2235
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.77645829482387e-05,
+      "loss": 0.2015,
+      "step": 2236
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7754804715794695e-05,
+      "loss": 0.145,
+      "step": 2237
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7745025271682293e-05,
+      "loss": 0.1307,
+      "step": 2238
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.773524462020288e-05,
+      "loss": 0.2475,
+      "step": 2239
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7725462765658358e-05,
+      "loss": 0.1731,
+      "step": 2240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.771567971235115e-05,
+      "loss": 0.1775,
+      "step": 2241
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.770589546458423e-05,
+      "loss": 0.1814,
+      "step": 2242
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7696110026661086e-05,
+      "loss": 0.2332,
+      "step": 2243
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.768632340288571e-05,
+      "loss": 0.1304,
+      "step": 2244
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7676535597562644e-05,
+      "loss": 0.2421,
+      "step": 2245
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.766674661499694e-05,
+      "loss": 0.2073,
+      "step": 2246
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.765695645949417e-05,
+      "loss": 0.1839,
+      "step": 2247
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7647165135360412e-05,
+      "loss": 0.1625,
+      "step": 2248
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.763737264690226e-05,
+      "loss": 0.2301,
+      "step": 2249
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7627578998426836e-05,
+      "loss": 0.1696,
+      "step": 2250
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7617784194241754e-05,
+      "loss": 0.2089,
+      "step": 2251
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7607988238655143e-05,
+      "loss": 0.1641,
+      "step": 2252
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.759819113597564e-05,
+      "loss": 0.2383,
+      "step": 2253
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7588392890512393e-05,
+      "loss": 0.158,
+      "step": 2254
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.757859350657503e-05,
+      "loss": 0.1898,
+      "step": 2255
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7568792988473702e-05,
+      "loss": 0.1848,
+      "step": 2256
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7558991340519054e-05,
+      "loss": 0.1869,
+      "step": 2257
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7549188567022224e-05,
+      "loss": 0.1887,
+      "step": 2258
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7539384672294837e-05,
+      "loss": 0.1568,
+      "step": 2259
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7529579660649034e-05,
+      "loss": 0.1225,
+      "step": 2260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.751977353639743e-05,
+      "loss": 0.2472,
+      "step": 2261
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.750996630385313e-05,
+      "loss": 0.26,
+      "step": 2262
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7500157967329733e-05,
+      "loss": 0.1798,
+      "step": 2263
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7490348531141312e-05,
+      "loss": 0.2355,
+      "step": 2264
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7480537999602444e-05,
+      "loss": 0.2299,
+      "step": 2265
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7470726377028168e-05,
+      "loss": 0.1736,
+      "step": 2266
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7460913667734012e-05,
+      "loss": 0.1886,
+      "step": 2267
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7451099876035986e-05,
+      "loss": 0.1412,
+      "step": 2268
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7441285006250565e-05,
+      "loss": 0.2006,
+      "step": 2269
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7431469062694698e-05,
+      "loss": 0.1777,
+      "step": 2270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7421652049685823e-05,
+      "loss": 0.1904,
+      "step": 2271
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7411833971541832e-05,
+      "loss": 0.2328,
+      "step": 2272
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.740201483258109e-05,
+      "loss": 0.2701,
+      "step": 2273
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7392194637122434e-05,
+      "loss": 0.1956,
+      "step": 2274
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7382373389485154e-05,
+      "loss": 0.135,
+      "step": 2275
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7372551093989022e-05,
+      "loss": 0.1899,
+      "step": 2276
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7362727754954248e-05,
+      "loss": 0.1898,
+      "step": 2277
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7352903376701518e-05,
+      "loss": 0.2172,
+      "step": 2278
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7343077963551965e-05,
+      "loss": 0.2276,
+      "step": 2279
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7333251519827184e-05,
+      "loss": 0.2368,
+      "step": 2280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.732342404984922e-05,
+      "loss": 0.2167,
+      "step": 2281
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7313595557940568e-05,
+      "loss": 0.2151,
+      "step": 2282
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7303766048424178e-05,
+      "loss": 0.1952,
+      "step": 2283
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7293935525623435e-05,
+      "loss": 0.1895,
+      "step": 2284
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.728410399386219e-05,
+      "loss": 0.1527,
+      "step": 2285
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7274271457464723e-05,
+      "loss": 0.1915,
+      "step": 2286
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7264437920755756e-05,
+      "loss": 0.2007,
+      "step": 2287
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.725460338806045e-05,
+      "loss": 0.1919,
+      "step": 2288
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7244767863704424e-05,
+      "loss": 0.2401,
+      "step": 2289
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.72349313520137e-05,
+      "loss": 0.1757,
+      "step": 2290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.722509385731476e-05,
+      "loss": 0.1957,
+      "step": 2291
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7215255383934517e-05,
+      "loss": 0.2144,
+      "step": 2292
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7205415936200296e-05,
+      "loss": 0.1409,
+      "step": 2293
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7195575518439868e-05,
+      "loss": 0.2039,
+      "step": 2294
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7185734134981416e-05,
+      "loss": 0.25,
+      "step": 2295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.717589179015357e-05,
+      "loss": 0.2602,
+      "step": 2296
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7166048488285362e-05,
+      "loss": 0.1698,
+      "step": 2297
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.715620423370625e-05,
+      "loss": 0.1849,
+      "step": 2298
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.714635903074612e-05,
+      "loss": 0.1887,
+      "step": 2299
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7136512883735265e-05,
+      "loss": 0.2314,
+      "step": 2300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7126665797004394e-05,
+      "loss": 0.1499,
+      "step": 2301
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7116817774884637e-05,
+      "loss": 0.1558,
+      "step": 2302
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.710696882170753e-05,
+      "loss": 0.2022,
+      "step": 2303
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7097118941805016e-05,
+      "loss": 0.1451,
+      "step": 2304
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7087268139509448e-05,
+      "loss": 0.1625,
+      "step": 2305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7077416419153584e-05,
+      "loss": 0.2105,
+      "step": 2306
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7067563785070598e-05,
+      "loss": 0.1828,
+      "step": 2307
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7057710241594046e-05,
+      "loss": 0.1687,
+      "step": 2308
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7047855793057894e-05,
+      "loss": 0.1666,
+      "step": 2309
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.70380004437965e-05,
+      "loss": 0.2254,
+      "step": 2310
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7028144198144636e-05,
+      "loss": 0.2547,
+      "step": 2311
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7018287060437444e-05,
+      "loss": 0.1829,
+      "step": 2312
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7008429035010474e-05,
+      "loss": 0.1721,
+      "step": 2313
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6998570126199664e-05,
+      "loss": 0.1826,
+      "step": 2314
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.698871033834134e-05,
+      "loss": 0.1352,
+      "step": 2315
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6978849675772204e-05,
+      "loss": 0.1823,
+      "step": 2316
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6968988142829362e-05,
+      "loss": 0.2073,
+      "step": 2317
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.695912574385029e-05,
+      "loss": 0.2687,
+      "step": 2318
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6949262483172848e-05,
+      "loss": 0.1864,
+      "step": 2319
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6939398365135274e-05,
+      "loss": 0.161,
+      "step": 2320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6929533394076185e-05,
+      "loss": 0.1798,
+      "step": 2321
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6919667574334576e-05,
+      "loss": 0.1613,
+      "step": 2322
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6909800910249796e-05,
+      "loss": 0.2336,
+      "step": 2323
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6899933406161603e-05,
+      "loss": 0.225,
+      "step": 2324
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.689006506641009e-05,
+      "loss": 0.1364,
+      "step": 2325
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6880195895335737e-05,
+      "loss": 0.2026,
+      "step": 2326
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6870325897279367e-05,
+      "loss": 0.1653,
+      "step": 2327
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.68604550765822e-05,
+      "loss": 0.1634,
+      "step": 2328
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.685058343758579e-05,
+      "loss": 0.1583,
+      "step": 2329
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6840710984632064e-05,
+      "loss": 0.1638,
+      "step": 2330
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6830837722063308e-05,
+      "loss": 0.1887,
+      "step": 2331
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6820963654222152e-05,
+      "loss": 0.219,
+      "step": 2332
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6811088785451595e-05,
+      "loss": 0.2566,
+      "step": 2333
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6801213120094977e-05,
+      "loss": 0.1706,
+      "step": 2334
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6791336662495993e-05,
+      "loss": 0.0899,
+      "step": 2335
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6781459416998694e-05,
+      "loss": 0.2554,
+      "step": 2336
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6771581387947468e-05,
+      "loss": 0.205,
+      "step": 2337
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.676170257968704e-05,
+      "loss": 0.1891,
+      "step": 2338
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6751822996562497e-05,
+      "loss": 0.2485,
+      "step": 2339
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6741942642919252e-05,
+      "loss": 0.2151,
+      "step": 2340
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6732061523103068e-05,
+      "loss": 0.2178,
+      "step": 2341
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6722179641460033e-05,
+      "loss": 0.1878,
+      "step": 2342
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.671229700233658e-05,
+      "loss": 0.1345,
+      "step": 2343
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6702413610079468e-05,
+      "loss": 0.1861,
+      "step": 2344
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6692529469035793e-05,
+      "loss": 0.1438,
+      "step": 2345
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6682644583552973e-05,
+      "loss": 0.1842,
+      "step": 2346
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6672758957978772e-05,
+      "loss": 0.2494,
+      "step": 2347
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6662872596661253e-05,
+      "loss": 0.206,
+      "step": 2348
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6652985503948815e-05,
+      "loss": 0.1988,
+      "step": 2349
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.664309768419019e-05,
+      "loss": 0.1962,
+      "step": 2350
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6633209141734413e-05,
+      "loss": 0.1196,
+      "step": 2351
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6623319880930836e-05,
+      "loss": 0.2048,
+      "step": 2352
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.661342990612914e-05,
+      "loss": 0.1667,
+      "step": 2353
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.660353922167932e-05,
+      "loss": 0.1767,
+      "step": 2354
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6593647831931666e-05,
+      "loss": 0.1239,
+      "step": 2355
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6583755741236796e-05,
+      "loss": 0.1867,
+      "step": 2356
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6573862953945632e-05,
+      "loss": 0.2183,
+      "step": 2357
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6563969474409392e-05,
+      "loss": 0.1937,
+      "step": 2358
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6554075306979607e-05,
+      "loss": 0.1713,
+      "step": 2359
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6544180456008117e-05,
+      "loss": 0.1909,
+      "step": 2360
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.653428492584705e-05,
+      "loss": 0.2261,
+      "step": 2361
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6524388720848837e-05,
+      "loss": 0.1825,
+      "step": 2362
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6514491845366218e-05,
+      "loss": 0.2005,
+      "step": 2363
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6504594303752206e-05,
+      "loss": 0.2176,
+      "step": 2364
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6494696100360115e-05,
+      "loss": 0.19,
+      "step": 2365
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6484797239543564e-05,
+      "loss": 0.1351,
+      "step": 2366
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6474897725656445e-05,
+      "loss": 0.1569,
+      "step": 2367
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6464997563052945e-05,
+      "loss": 0.2325,
+      "step": 2368
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6455096756087528e-05,
+      "loss": 0.1979,
+      "step": 2369
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6445195309114953e-05,
+      "loss": 0.1583,
+      "step": 2370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6435293226490253e-05,
+      "loss": 0.185,
+      "step": 2371
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6425390512568747e-05,
+      "loss": 0.1745,
+      "step": 2372
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6415487171706016e-05,
+      "loss": 0.1396,
+      "step": 2373
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6405583208257935e-05,
+      "loss": 0.1664,
+      "step": 2374
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.639567862658065e-05,
+      "loss": 0.2125,
+      "step": 2375
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.638577343103056e-05,
+      "loss": 0.2231,
+      "step": 2376
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.637586762596436e-05,
+      "loss": 0.122,
+      "step": 2377
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6365961215739002e-05,
+      "loss": 0.1779,
+      "step": 2378
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6356054204711704e-05,
+      "loss": 0.1873,
+      "step": 2379
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.634614659723994e-05,
+      "loss": 0.1743,
+      "step": 2380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6336238397681462e-05,
+      "loss": 0.2164,
+      "step": 2381
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6326329610394272e-05,
+      "loss": 0.1882,
+      "step": 2382
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6316420239736636e-05,
+      "loss": 0.1792,
+      "step": 2383
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6306510290067064e-05,
+      "loss": 0.1356,
+      "step": 2384
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6296599765744344e-05,
+      "loss": 0.1547,
+      "step": 2385
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6286688671127496e-05,
+      "loss": 0.2027,
+      "step": 2386
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.62767770105758e-05,
+      "loss": 0.1668,
+      "step": 2387
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.626686478844878e-05,
+      "loss": 0.1825,
+      "step": 2388
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.625695200910622e-05,
+      "loss": 0.1652,
+      "step": 2389
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6247038676908127e-05,
+      "loss": 0.1528,
+      "step": 2390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6237124796214765e-05,
+      "loss": 0.1229,
+      "step": 2391
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.622721037138665e-05,
+      "loss": 0.1716,
+      "step": 2392
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.621729540678451e-05,
+      "loss": 0.1585,
+      "step": 2393
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6207379906769332e-05,
+      "loss": 0.1542,
+      "step": 2394
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6197463875702327e-05,
+      "loss": 0.1724,
+      "step": 2395
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6187547317944953e-05,
+      "loss": 0.1795,
+      "step": 2396
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6177630237858884e-05,
+      "loss": 0.1726,
+      "step": 2397
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6167712639806036e-05,
+      "loss": 0.1329,
+      "step": 2398
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.615779452814854e-05,
+      "loss": 0.1305,
+      "step": 2399
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6147875907248774e-05,
+      "loss": 0.1588,
+      "step": 2400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.613795678146931e-05,
+      "loss": 0.1747,
+      "step": 2401
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6128037155172966e-05,
+      "loss": 0.1358,
+      "step": 2402
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.611811703272278e-05,
+      "loss": 0.1599,
+      "step": 2403
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.610819641848199e-05,
+      "loss": 0.1884,
+      "step": 2404
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6098275316814066e-05,
+      "loss": 0.1812,
+      "step": 2405
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6088353732082687e-05,
+      "loss": 0.1982,
+      "step": 2406
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6078431668651745e-05,
+      "loss": 0.187,
+      "step": 2407
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6068509130885345e-05,
+      "loss": 0.1959,
+      "step": 2408
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.605858612314779e-05,
+      "loss": 0.1902,
+      "step": 2409
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.604866264980361e-05,
+      "loss": 0.1774,
+      "step": 2410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.603873871521752e-05,
+      "loss": 0.1308,
+      "step": 2411
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.602881432375445e-05,
+      "loss": 0.1531,
+      "step": 2412
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.601888947977952e-05,
+      "loss": 0.1645,
+      "step": 2413
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6008964187658063e-05,
+      "loss": 0.1577,
+      "step": 2414
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.59990384517556e-05,
+      "loss": 0.1812,
+      "step": 2415
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.598911227643784e-05,
+      "loss": 0.2372,
+      "step": 2416
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.597918566607071e-05,
+      "loss": 0.189,
+      "step": 2417
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5969258625020304e-05,
+      "loss": 0.2053,
+      "step": 2418
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5959331157652916e-05,
+      "loss": 0.1992,
+      "step": 2419
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5949403268335015e-05,
+      "loss": 0.1782,
+      "step": 2420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.593947496143328e-05,
+      "loss": 0.1773,
+      "step": 2421
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.592954624131456e-05,
+      "loss": 0.1929,
+      "step": 2422
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.591961711234587e-05,
+      "loss": 0.2001,
+      "step": 2423
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5909687578894438e-05,
+      "loss": 0.1548,
+      "step": 2424
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5899757645327645e-05,
+      "loss": 0.2482,
+      "step": 2425
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5889827316013053e-05,
+      "loss": 0.1822,
+      "step": 2426
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.58798965953184e-05,
+      "loss": 0.1826,
+      "step": 2427
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5869965487611606e-05,
+      "loss": 0.2009,
+      "step": 2428
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5860033997260743e-05,
+      "loss": 0.1203,
+      "step": 2429
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.585010212863406e-05,
+      "loss": 0.1962,
+      "step": 2430
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5840169886099973e-05,
+      "loss": 0.2614,
+      "step": 2431
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5830237274027065e-05,
+      "loss": 0.1631,
+      "step": 2432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5820304296784078e-05,
+      "loss": 0.1402,
+      "step": 2433
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5810370958739908e-05,
+      "loss": 0.1559,
+      "step": 2434
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5800437264263627e-05,
+      "loss": 0.2033,
+      "step": 2435
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5790503217724448e-05,
+      "loss": 0.1786,
+      "step": 2436
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5780568823491742e-05,
+      "loss": 0.2022,
+      "step": 2437
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5770634085935037e-05,
+      "loss": 0.2188,
+      "step": 2438
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5760699009424018e-05,
+      "loss": 0.1474,
+      "step": 2439
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.57507635983285e-05,
+      "loss": 0.149,
+      "step": 2440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5740827857018457e-05,
+      "loss": 0.1668,
+      "step": 2441
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5730891789864018e-05,
+      "loss": 0.2116,
+      "step": 2442
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5720955401235436e-05,
+      "loss": 0.1111,
+      "step": 2443
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5711018695503122e-05,
+      "loss": 0.1302,
+      "step": 2444
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5701081677037608e-05,
+      "loss": 0.1703,
+      "step": 2445
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.569114435020958e-05,
+      "loss": 0.1563,
+      "step": 2446
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5681206719389857e-05,
+      "loss": 0.1521,
+      "step": 2447
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5671268788949387e-05,
+      "loss": 0.1003,
+      "step": 2448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.566133056325925e-05,
+      "loss": 0.085,
+      "step": 2449
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5651392046690664e-05,
+      "loss": 0.0908,
+      "step": 2450
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5641453243614957e-05,
+      "loss": 0.1309,
+      "step": 2451
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.56315141584036e-05,
+      "loss": 0.1161,
+      "step": 2452
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5621574795428185e-05,
+      "loss": 0.1287,
+      "step": 2453
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5611635159060425e-05,
+      "loss": 0.1258,
+      "step": 2454
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5601695253672137e-05,
+      "loss": 0.1568,
+      "step": 2455
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.559175508363529e-05,
+      "loss": 0.1266,
+      "step": 2456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5581814653321946e-05,
+      "loss": 0.1238,
+      "step": 2457
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.557187396710428e-05,
+      "loss": 0.107,
+      "step": 2458
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5561933029354585e-05,
+      "loss": 0.1023,
+      "step": 2459
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5551991844445277e-05,
+      "loss": 0.1353,
+      "step": 2460
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.554205041674886e-05,
+      "loss": 0.1248,
+      "step": 2461
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5532108750637957e-05,
+      "loss": 0.1125,
+      "step": 2462
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5522166850485283e-05,
+      "loss": 0.175,
+      "step": 2463
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.551222472066368e-05,
+      "loss": 0.1172,
+      "step": 2464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.550228236554607e-05,
+      "loss": 0.0822,
+      "step": 2465
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5492339789505475e-05,
+      "loss": 0.1132,
+      "step": 2466
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5482396996915033e-05,
+      "loss": 0.1011,
+      "step": 2467
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5472453992147953e-05,
+      "loss": 0.1404,
+      "step": 2468
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5462510779577558e-05,
+      "loss": 0.1106,
+      "step": 2469
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.545256736357724e-05,
+      "loss": 0.1662,
+      "step": 2470
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54426237485205e-05,
+      "loss": 0.1351,
+      "step": 2471
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5432679938780928e-05,
+      "loss": 0.1101,
+      "step": 2472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5422735938732178e-05,
+      "loss": 0.1096,
+      "step": 2473
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5412791752748014e-05,
+      "loss": 0.1703,
+      "step": 2474
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5402847385202256e-05,
+      "loss": 0.1016,
+      "step": 2475
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5392902840468827e-05,
+      "loss": 0.1217,
+      "step": 2476
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5382958122921706e-05,
+      "loss": 0.1543,
+      "step": 2477
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.537301323693498e-05,
+      "loss": 0.1242,
+      "step": 2478
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536306818688277e-05,
+      "loss": 0.1178,
+      "step": 2479
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5353122977139292e-05,
+      "loss": 0.1237,
+      "step": 2480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5343177612078845e-05,
+      "loss": 0.1444,
+      "step": 2481
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5333232096075764e-05,
+      "loss": 0.1,
+      "step": 2482
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5323286433504477e-05,
+      "loss": 0.1173,
+      "step": 2483
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5313340628739456e-05,
+      "loss": 0.0958,
+      "step": 2484
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5303394686155255e-05,
+      "loss": 0.1503,
+      "step": 2485
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5293448610126475e-05,
+      "loss": 0.1445,
+      "step": 2486
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5283502405027783e-05,
+      "loss": 0.1303,
+      "step": 2487
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5273556075233895e-05,
+      "loss": 0.1596,
+      "step": 2488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.52636096251196e-05,
+      "loss": 0.1363,
+      "step": 2489
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5253663059059714e-05,
+      "loss": 0.152,
+      "step": 2490
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5243716381429117e-05,
+      "loss": 0.1471,
+      "step": 2491
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.523376959660275e-05,
+      "loss": 0.1276,
+      "step": 2492
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5223822708955577e-05,
+      "loss": 0.115,
+      "step": 2493
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5213875722862622e-05,
+      "loss": 0.1217,
+      "step": 2494
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.520392864269895e-05,
+      "loss": 0.128,
+      "step": 2495
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5193981472839673e-05,
+      "loss": 0.1469,
+      "step": 2496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5184034217659929e-05,
+      "loss": 0.093,
+      "step": 2497
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5174086881534897e-05,
+      "loss": 0.0922,
+      "step": 2498
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.516413946883981e-05,
+      "loss": 0.13,
+      "step": 2499
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5154191983949908e-05,
+      "loss": 0.105,
+      "step": 2500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5144244431240483e-05,
+      "loss": 0.0763,
+      "step": 2501
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5134296815086838e-05,
+      "loss": 0.1293,
+      "step": 2502
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5124349139864326e-05,
+      "loss": 0.1067,
+      "step": 2503
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5114401409948312e-05,
+      "loss": 0.138,
+      "step": 2504
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.510445362971418e-05,
+      "loss": 0.1641,
+      "step": 2505
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5094505803537354e-05,
+      "loss": 0.1314,
+      "step": 2506
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5084557935793264e-05,
+      "loss": 0.1285,
+      "step": 2507
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5074610030857365e-05,
+      "loss": 0.0903,
+      "step": 2508
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5064662093105119e-05,
+      "loss": 0.1092,
+      "step": 2509
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5054714126912012e-05,
+      "loss": 0.1289,
+      "step": 2510
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5044766136653547e-05,
+      "loss": 0.0962,
+      "step": 2511
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5034818126705219e-05,
+      "loss": 0.1747,
+      "step": 2512
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.502487010144255e-05,
+      "loss": 0.1445,
+      "step": 2513
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5014922065241062e-05,
+      "loss": 0.1099,
+      "step": 2514
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5004974022476278e-05,
+      "loss": 0.105,
+      "step": 2515
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4995025977523722e-05,
+      "loss": 0.1153,
+      "step": 2516
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4985077934758944e-05,
+      "loss": 0.1241,
+      "step": 2517
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4975129898557453e-05,
+      "loss": 0.1469,
+      "step": 2518
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4965181873294784e-05,
+      "loss": 0.1365,
+      "step": 2519
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4955233863346455e-05,
+      "loss": 0.1275,
+      "step": 2520
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4945285873087992e-05,
+      "loss": 0.1075,
+      "step": 2521
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4935337906894885e-05,
+      "loss": 0.1043,
+      "step": 2522
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4925389969142641e-05,
+      "loss": 0.1238,
+      "step": 2523
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4915442064206737e-05,
+      "loss": 0.1386,
+      "step": 2524
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4905494196462651e-05,
+      "loss": 0.1185,
+      "step": 2525
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4895546370285822e-05,
+      "loss": 0.1427,
+      "step": 2526
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4885598590051696e-05,
+      "loss": 0.1032,
+      "step": 2527
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4875650860135677e-05,
+      "loss": 0.1683,
+      "step": 2528
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4865703184913168e-05,
+      "loss": 0.1329,
+      "step": 2529
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4855755568759522e-05,
+      "loss": 0.1283,
+      "step": 2530
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.484580801605009e-05,
+      "loss": 0.1016,
+      "step": 2531
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4835860531160192e-05,
+      "loss": 0.0965,
+      "step": 2532
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4825913118465102e-05,
+      "loss": 0.1094,
+      "step": 2533
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4815965782340075e-05,
+      "loss": 0.0989,
+      "step": 2534
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4806018527160327e-05,
+      "loss": 0.0964,
+      "step": 2535
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4796071357301051e-05,
+      "loss": 0.1069,
+      "step": 2536
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4786124277137379e-05,
+      "loss": 0.1483,
+      "step": 2537
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4776177291044425e-05,
+      "loss": 0.093,
+      "step": 2538
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.476623040339725e-05,
+      "loss": 0.1072,
+      "step": 2539
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4756283618570883e-05,
+      "loss": 0.104,
+      "step": 2540
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4746336940940283e-05,
+      "loss": 0.1247,
+      "step": 2541
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4736390374880402e-05,
+      "loss": 0.1037,
+      "step": 2542
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.47264439247661e-05,
+      "loss": 0.1341,
+      "step": 2543
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.471649759497222e-05,
+      "loss": 0.143,
+      "step": 2544
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4706551389873526e-05,
+      "loss": 0.1611,
+      "step": 2545
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4696605313844753e-05,
+      "loss": 0.1586,
+      "step": 2546
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4686659371260549e-05,
+      "loss": 0.1402,
+      "step": 2547
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.467671356649553e-05,
+      "loss": 0.1248,
+      "step": 2548
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4666767903924238e-05,
+      "loss": 0.1343,
+      "step": 2549
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4656822387921163e-05,
+      "loss": 0.0917,
+      "step": 2550
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4646877022860707e-05,
+      "loss": 0.1231,
+      "step": 2551
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4636931813117237e-05,
+      "loss": 0.131,
+      "step": 2552
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4626986763065027e-05,
+      "loss": 0.1286,
+      "step": 2553
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4617041877078298e-05,
+      "loss": 0.1247,
+      "step": 2554
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4607097159531178e-05,
+      "loss": 0.1344,
+      "step": 2555
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.459715261479775e-05,
+      "loss": 0.1311,
+      "step": 2556
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4587208247251992e-05,
+      "loss": 0.0926,
+      "step": 2557
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4577264061267824e-05,
+      "loss": 0.105,
+      "step": 2558
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4567320061219074e-05,
+      "loss": 0.1166,
+      "step": 2559
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4557376251479501e-05,
+      "loss": 0.1367,
+      "step": 2560
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4547432636422764e-05,
+      "loss": 0.1119,
+      "step": 2561
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4537489220422446e-05,
+      "loss": 0.1337,
+      "step": 2562
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4527546007852046e-05,
+      "loss": 0.1035,
+      "step": 2563
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4517603003084966e-05,
+      "loss": 0.1266,
+      "step": 2564
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4507660210494525e-05,
+      "loss": 0.1272,
+      "step": 2565
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4497717634453932e-05,
+      "loss": 0.1179,
+      "step": 2566
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4487775279336323e-05,
+      "loss": 0.0999,
+      "step": 2567
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4477833149514716e-05,
+      "loss": 0.1437,
+      "step": 2568
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4467891249362047e-05,
+      "loss": 0.1176,
+      "step": 2569
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.445794958325114e-05,
+      "loss": 0.1317,
+      "step": 2570
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4448008155554724e-05,
+      "loss": 0.1021,
+      "step": 2571
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4438066970645412e-05,
+      "loss": 0.1132,
+      "step": 2572
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4428126032895723e-05,
+      "loss": 0.138,
+      "step": 2573
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4418185346678055e-05,
+      "loss": 0.112,
+      "step": 2574
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4408244916364712e-05,
+      "loss": 0.105,
+      "step": 2575
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4398304746327864e-05,
+      "loss": 0.1208,
+      "step": 2576
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4388364840939585e-05,
+      "loss": 0.1124,
+      "step": 2577
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4378425204571817e-05,
+      "loss": 0.1379,
+      "step": 2578
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4368485841596406e-05,
+      "loss": 0.1765,
+      "step": 2579
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4358546756385046e-05,
+      "loss": 0.1387,
+      "step": 2580
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4348607953309344e-05,
+      "loss": 0.1007,
+      "step": 2581
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.433866943674075e-05,
+      "loss": 0.0949,
+      "step": 2582
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4328731211050616e-05,
+      "loss": 0.112,
+      "step": 2583
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4318793280610143e-05,
+      "loss": 0.1566,
+      "step": 2584
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4308855649790424e-05,
+      "loss": 0.1514,
+      "step": 2585
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4298918322962398e-05,
+      "loss": 0.1466,
+      "step": 2586
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4288981304496887e-05,
+      "loss": 0.1057,
+      "step": 2587
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4279044598764567e-05,
+      "loss": 0.0957,
+      "step": 2588
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.426910821013599e-05,
+      "loss": 0.1213,
+      "step": 2589
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4259172142981544e-05,
+      "loss": 0.0675,
+      "step": 2590
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4249236401671505e-05,
+      "loss": 0.1534,
+      "step": 2591
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4239300990575988e-05,
+      "loss": 0.1135,
+      "step": 2592
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4229365914064962e-05,
+      "loss": 0.1394,
+      "step": 2593
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.421943117650826e-05,
+      "loss": 0.1417,
+      "step": 2594
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4209496782275555e-05,
+      "loss": 0.114,
+      "step": 2595
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4199562735736377e-05,
+      "loss": 0.1583,
+      "step": 2596
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4189629041260093e-05,
+      "loss": 0.0836,
+      "step": 2597
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4179695703215927e-05,
+      "loss": 0.1649,
+      "step": 2598
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4169762725972936e-05,
+      "loss": 0.086,
+      "step": 2599
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.415983011390003e-05,
+      "loss": 0.0821,
+      "step": 2600
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4149897871365942e-05,
+      "loss": 0.1568,
+      "step": 2601
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4139966002739263e-05,
+      "loss": 0.1035,
+      "step": 2602
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4130034512388395e-05,
+      "loss": 0.0847,
+      "step": 2603
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4120103404681602e-05,
+      "loss": 0.1565,
+      "step": 2604
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.411017268398695e-05,
+      "loss": 0.1079,
+      "step": 2605
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.410024235467236e-05,
+      "loss": 0.0974,
+      "step": 2606
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4090312421105563e-05,
+      "loss": 0.1347,
+      "step": 2607
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.408038288765413e-05,
+      "loss": 0.1118,
+      "step": 2608
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4070453758685446e-05,
+      "loss": 0.0752,
+      "step": 2609
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4060525038566723e-05,
+      "loss": 0.1353,
+      "step": 2610
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4050596731664988e-05,
+      "loss": 0.1307,
+      "step": 2611
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4040668842347093e-05,
+      "loss": 0.0941,
+      "step": 2612
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.40307413749797e-05,
+      "loss": 0.1159,
+      "step": 2613
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4020814333929295e-05,
+      "loss": 0.1399,
+      "step": 2614
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4010887723562158e-05,
+      "loss": 0.1339,
+      "step": 2615
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4000961548244406e-05,
+      "loss": 0.1092,
+      "step": 2616
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3991035812341938e-05,
+      "loss": 0.1003,
+      "step": 2617
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3981110520220485e-05,
+      "loss": 0.0974,
+      "step": 2618
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3971185676245552e-05,
+      "loss": 0.1359,
+      "step": 2619
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3961261284782485e-05,
+      "loss": 0.1333,
+      "step": 2620
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3951337350196391e-05,
+      "loss": 0.1089,
+      "step": 2621
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3941413876852209e-05,
+      "loss": 0.1416,
+      "step": 2622
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3931490869114659e-05,
+      "loss": 0.1193,
+      "step": 2623
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3921568331348254e-05,
+      "loss": 0.118,
+      "step": 2624
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3911646267917317e-05,
+      "loss": 0.1462,
+      "step": 2625
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3901724683185935e-05,
+      "loss": 0.1115,
+      "step": 2626
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3891803581518011e-05,
+      "loss": 0.0815,
+      "step": 2627
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.388188296727722e-05,
+      "loss": 0.1022,
+      "step": 2628
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3871962844827034e-05,
+      "loss": 0.1376,
+      "step": 2629
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3862043218530692e-05,
+      "loss": 0.1444,
+      "step": 2630
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3852124092751232e-05,
+      "loss": 0.088,
+      "step": 2631
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3842205471851458e-05,
+      "loss": 0.1322,
+      "step": 2632
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3832287360193967e-05,
+      "loss": 0.1364,
+      "step": 2633
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3822369762141115e-05,
+      "loss": 0.1049,
+      "step": 2634
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3812452682055052e-05,
+      "loss": 0.1224,
+      "step": 2635
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3802536124297674e-05,
+      "loss": 0.0983,
+      "step": 2636
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3792620093230674e-05,
+      "loss": 0.0956,
+      "step": 2637
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3782704593215491e-05,
+      "loss": 0.1504,
+      "step": 2638
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3772789628613358e-05,
+      "loss": 0.1425,
+      "step": 2639
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3762875203785234e-05,
+      "loss": 0.0957,
+      "step": 2640
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3752961323091878e-05,
+      "loss": 0.0964,
+      "step": 2641
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3743047990893783e-05,
+      "loss": 0.0838,
+      "step": 2642
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3733135211551222e-05,
+      "loss": 0.1163,
+      "step": 2643
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3723222989424203e-05,
+      "loss": 0.1404,
+      "step": 2644
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3713311328872508e-05,
+      "loss": 0.1278,
+      "step": 2645
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3703400234255657e-05,
+      "loss": 0.1753,
+      "step": 2646
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3693489709932938e-05,
+      "loss": 0.1214,
+      "step": 2647
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3683579760263368e-05,
+      "loss": 0.1025,
+      "step": 2648
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3673670389605735e-05,
+      "loss": 0.1063,
+      "step": 2649
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3663761602318542e-05,
+      "loss": 0.1299,
+      "step": 2650
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3653853402760063e-05,
+      "loss": 0.1214,
+      "step": 2651
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3643945795288299e-05,
+      "loss": 0.1486,
+      "step": 2652
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3634038784260995e-05,
+      "loss": 0.1297,
+      "step": 2653
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3624132374035641e-05,
+      "loss": 0.107,
+      "step": 2654
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3614226568969441e-05,
+      "loss": 0.1287,
+      "step": 2655
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3604321373419357e-05,
+      "loss": 0.139,
+      "step": 2656
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3594416791742065e-05,
+      "loss": 0.1282,
+      "step": 2657
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3584512828293984e-05,
+      "loss": 0.0951,
+      "step": 2658
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3574609487431254e-05,
+      "loss": 0.1164,
+      "step": 2659
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3564706773509748e-05,
+      "loss": 0.1332,
+      "step": 2660
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3554804690885046e-05,
+      "loss": 0.0801,
+      "step": 2661
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3544903243912473e-05,
+      "loss": 0.1083,
+      "step": 2662
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3535002436947056e-05,
+      "loss": 0.1198,
+      "step": 2663
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3525102274343557e-05,
+      "loss": 0.1064,
+      "step": 2664
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3515202760456437e-05,
+      "loss": 0.1303,
+      "step": 2665
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3505303899639886e-05,
+      "loss": 0.1227,
+      "step": 2666
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3495405696247798e-05,
+      "loss": 0.1221,
+      "step": 2667
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.348550815463379e-05,
+      "loss": 0.1543,
+      "step": 2668
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3475611279151162e-05,
+      "loss": 0.1027,
+      "step": 2669
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3465715074152954e-05,
+      "loss": 0.1596,
+      "step": 2670
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3455819543991885e-05,
+      "loss": 0.1189,
+      "step": 2671
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3445924693020396e-05,
+      "loss": 0.1002,
+      "step": 2672
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3436030525590612e-05,
+      "loss": 0.1557,
+      "step": 2673
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3426137046054375e-05,
+      "loss": 0.0841,
+      "step": 2674
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3416244258763206e-05,
+      "loss": 0.1431,
+      "step": 2675
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3406352168068336e-05,
+      "loss": 0.125,
+      "step": 2676
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3396460778320683e-05,
+      "loss": 0.1435,
+      "step": 2677
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3386570093870864e-05,
+      "loss": 0.1258,
+      "step": 2678
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3376680119069168e-05,
+      "loss": 0.0806,
+      "step": 2679
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3366790858265597e-05,
+      "loss": 0.1279,
+      "step": 2680
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3356902315809813e-05,
+      "loss": 0.1104,
+      "step": 2681
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3347014496051182e-05,
+      "loss": 0.1219,
+      "step": 2682
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.333712740333875e-05,
+      "loss": 0.1075,
+      "step": 2683
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3327241042021227e-05,
+      "loss": 0.1541,
+      "step": 2684
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3317355416447026e-05,
+      "loss": 0.0985,
+      "step": 2685
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.330747053096421e-05,
+      "loss": 0.1051,
+      "step": 2686
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3297586389920536e-05,
+      "loss": 0.1118,
+      "step": 2687
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3287702997663422e-05,
+      "loss": 0.1046,
+      "step": 2688
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3277820358539971e-05,
+      "loss": 0.1385,
+      "step": 2689
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3267938476896935e-05,
+      "loss": 0.1076,
+      "step": 2690
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3258057357080749e-05,
+      "loss": 0.1082,
+      "step": 2691
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3248177003437504e-05,
+      "loss": 0.1201,
+      "step": 2692
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3238297420312965e-05,
+      "loss": 0.0773,
+      "step": 2693
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3228418612052536e-05,
+      "loss": 0.1467,
+      "step": 2694
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3218540583001305e-05,
+      "loss": 0.1334,
+      "step": 2695
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3208663337504005e-05,
+      "loss": 0.1005,
+      "step": 2696
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3198786879905025e-05,
+      "loss": 0.1436,
+      "step": 2697
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3188911214548406e-05,
+      "loss": 0.1036,
+      "step": 2698
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3179036345777854e-05,
+      "loss": 0.1463,
+      "step": 2699
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3169162277936698e-05,
+      "loss": 0.1427,
+      "step": 2700
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3159289015367938e-05,
+      "loss": 0.1054,
+      "step": 2701
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.314941656241421e-05,
+      "loss": 0.0995,
+      "step": 2702
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3139544923417806e-05,
+      "loss": 0.1302,
+      "step": 2703
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3129674102720635e-05,
+      "loss": 0.1365,
+      "step": 2704
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3119804104664272e-05,
+      "loss": 0.1134,
+      "step": 2705
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.310993493358991e-05,
+      "loss": 0.1382,
+      "step": 2706
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3100066593838401e-05,
+      "loss": 0.1594,
+      "step": 2707
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3090199089750203e-05,
+      "loss": 0.0907,
+      "step": 2708
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3080332425665433e-05,
+      "loss": 0.1105,
+      "step": 2709
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.307046660592382e-05,
+      "loss": 0.1161,
+      "step": 2710
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.306060163486473e-05,
+      "loss": 0.1231,
+      "step": 2711
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3050737516827154e-05,
+      "loss": 0.1119,
+      "step": 2712
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3040874256149709e-05,
+      "loss": 0.1064,
+      "step": 2713
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.303101185717064e-05,
+      "loss": 0.1208,
+      "step": 2714
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3021150324227797e-05,
+      "loss": 0.119,
+      "step": 2715
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3011289661658665e-05,
+      "loss": 0.1276,
+      "step": 2716
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3001429873800335e-05,
+      "loss": 0.1231,
+      "step": 2717
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2991570964989528e-05,
+      "loss": 0.1054,
+      "step": 2718
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2981712939562558e-05,
+      "loss": 0.1415,
+      "step": 2719
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2971855801855368e-05,
+      "loss": 0.1398,
+      "step": 2720
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2961999556203497e-05,
+      "loss": 0.0997,
+      "step": 2721
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2952144206942109e-05,
+      "loss": 0.0857,
+      "step": 2722
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2942289758405953e-05,
+      "loss": 0.1741,
+      "step": 2723
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2932436214929405e-05,
+      "loss": 0.155,
+      "step": 2724
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2922583580846414e-05,
+      "loss": 0.0858,
+      "step": 2725
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2912731860490557e-05,
+      "loss": 0.1153,
+      "step": 2726
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2902881058194986e-05,
+      "loss": 0.1244,
+      "step": 2727
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2893031178292477e-05,
+      "loss": 0.1032,
+      "step": 2728
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2883182225115365e-05,
+      "loss": 0.1216,
+      "step": 2729
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2873334202995609e-05,
+      "loss": 0.1241,
+      "step": 2730
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.286348711626474e-05,
+      "loss": 0.121,
+      "step": 2731
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2853640969253887e-05,
+      "loss": 0.1471,
+      "step": 2732
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2843795766293755e-05,
+      "loss": 0.1195,
+      "step": 2733
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2833951511714645e-05,
+      "loss": 0.0853,
+      "step": 2734
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2824108209846434e-05,
+      "loss": 0.1198,
+      "step": 2735
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2814265865018587e-05,
+      "loss": 0.1136,
+      "step": 2736
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2804424481560138e-05,
+      "loss": 0.1311,
+      "step": 2737
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2794584063799712e-05,
+      "loss": 0.0902,
+      "step": 2738
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2784744616065486e-05,
+      "loss": 0.1347,
+      "step": 2739
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2774906142685241e-05,
+      "loss": 0.084,
+      "step": 2740
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.27650686479863e-05,
+      "loss": 0.1257,
+      "step": 2741
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2755232136295577e-05,
+      "loss": 0.1056,
+      "step": 2742
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2745396611939551e-05,
+      "loss": 0.1091,
+      "step": 2743
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2735562079244247e-05,
+      "loss": 0.1225,
+      "step": 2744
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2725728542535281e-05,
+      "loss": 0.0965,
+      "step": 2745
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.271589600613781e-05,
+      "loss": 0.0866,
+      "step": 2746
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2706064474376567e-05,
+      "loss": 0.1151,
+      "step": 2747
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2696233951575823e-05,
+      "loss": 0.146,
+      "step": 2748
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2686404442059435e-05,
+      "loss": 0.1573,
+      "step": 2749
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.267657595015078e-05,
+      "loss": 0.1104,
+      "step": 2750
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2666748480172817e-05,
+      "loss": 0.0695,
+      "step": 2751
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2656922036448032e-05,
+      "loss": 0.0917,
+      "step": 2752
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2647096623298485e-05,
+      "loss": 0.1083,
+      "step": 2753
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2637272245045751e-05,
+      "loss": 0.1415,
+      "step": 2754
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2627448906010979e-05,
+      "loss": 0.1443,
+      "step": 2755
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2617626610514842e-05,
+      "loss": 0.1041,
+      "step": 2756
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2607805362877571e-05,
+      "loss": 0.1381,
+      "step": 2757
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2597985167418912e-05,
+      "loss": 0.1055,
+      "step": 2758
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2588166028458174e-05,
+      "loss": 0.1199,
+      "step": 2759
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.257834795031418e-05,
+      "loss": 0.0682,
+      "step": 2760
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2568530937305306e-05,
+      "loss": 0.084,
+      "step": 2761
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2558714993749439e-05,
+      "loss": 0.151,
+      "step": 2762
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2548900123964018e-05,
+      "loss": 0.1296,
+      "step": 2763
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2539086332265987e-05,
+      "loss": 0.1301,
+      "step": 2764
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2529273622971833e-05,
+      "loss": 0.107,
+      "step": 2765
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2519462000397555e-05,
+      "loss": 0.0796,
+      "step": 2766
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.250965146885869e-05,
+      "loss": 0.1328,
+      "step": 2767
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2499842032670273e-05,
+      "loss": 0.178,
+      "step": 2768
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2490033696146873e-05,
+      "loss": 0.1176,
+      "step": 2769
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2480226463602573e-05,
+      "loss": 0.1083,
+      "step": 2770
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.247042033935097e-05,
+      "loss": 0.1412,
+      "step": 2771
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2460615327705165e-05,
+      "loss": 0.0886,
+      "step": 2772
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2450811432977777e-05,
+      "loss": 0.1751,
+      "step": 2773
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2441008659480949e-05,
+      "loss": 0.1405,
+      "step": 2774
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2431207011526297e-05,
+      "loss": 0.1212,
+      "step": 2775
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2421406493424972e-05,
+      "loss": 0.1048,
+      "step": 2776
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2411607109487608e-05,
+      "loss": 0.1101,
+      "step": 2777
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.240180886402436e-05,
+      "loss": 0.1555,
+      "step": 2778
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2392011761344857e-05,
+      "loss": 0.1063,
+      "step": 2779
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2382215805758249e-05,
+      "loss": 0.1325,
+      "step": 2780
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2372421001573165e-05,
+      "loss": 0.1588,
+      "step": 2781
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2362627353097744e-05,
+      "loss": 0.1278,
+      "step": 2782
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2352834864639592e-05,
+      "loss": 0.1505,
+      "step": 2783
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2343043540505835e-05,
+      "loss": 0.0929,
+      "step": 2784
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2333253385003057e-05,
+      "loss": 0.1042,
+      "step": 2785
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2323464402437359e-05,
+      "loss": 0.1404,
+      "step": 2786
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.231367659711429e-05,
+      "loss": 0.1381,
+      "step": 2787
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2303889973338922e-05,
+      "loss": 0.148,
+      "step": 2788
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.229410453541577e-05,
+      "loss": 0.1134,
+      "step": 2789
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2284320287648852e-05,
+      "loss": 0.1218,
+      "step": 2790
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2274537234341645e-05,
+      "loss": 0.1103,
+      "step": 2791
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2264755379797124e-05,
+      "loss": 0.114,
+      "step": 2792
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.225497472831771e-05,
+      "loss": 0.0707,
+      "step": 2793
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2245195284205314e-05,
+      "loss": 0.1133,
+      "step": 2794
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2235417051761302e-05,
+      "loss": 0.1387,
+      "step": 2795
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2225640035286527e-05,
+      "loss": 0.1086,
+      "step": 2796
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2215864239081281e-05,
+      "loss": 0.0863,
+      "step": 2797
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2206089667445338e-05,
+      "loss": 0.0877,
+      "step": 2798
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.219631632467793e-05,
+      "loss": 0.0995,
+      "step": 2799
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2186544215077747e-05,
+      "loss": 0.0996,
+      "step": 2800
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2176773342942928e-05,
+      "loss": 0.1132,
+      "step": 2801
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2167003712571094e-05,
+      "loss": 0.0873,
+      "step": 2802
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2157235328259279e-05,
+      "loss": 0.1145,
+      "step": 2803
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2147468194304006e-05,
+      "loss": 0.0809,
+      "step": 2804
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2137702315001232e-05,
+      "loss": 0.1512,
+      "step": 2805
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2127937694646362e-05,
+      "loss": 0.114,
+      "step": 2806
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2118174337534255e-05,
+      "loss": 0.1648,
+      "step": 2807
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2108412247959195e-05,
+      "loss": 0.1229,
+      "step": 2808
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2098651430214936e-05,
+      "loss": 0.1291,
+      "step": 2809
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2088891888594649e-05,
+      "loss": 0.0927,
+      "step": 2810
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2079133627390964e-05,
+      "loss": 0.123,
+      "step": 2811
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2069376650895922e-05,
+      "loss": 0.139,
+      "step": 2812
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.205962096340103e-05,
+      "loss": 0.1139,
+      "step": 2813
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2049866569197201e-05,
+      "loss": 0.1398,
+      "step": 2814
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2040113472574796e-05,
+      "loss": 0.1487,
+      "step": 2815
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2030361677823595e-05,
+      "loss": 0.0934,
+      "step": 2816
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.202061118923282e-05,
+      "loss": 0.0885,
+      "step": 2817
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2010862011091093e-05,
+      "loss": 0.1106,
+      "step": 2818
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2001114147686487e-05,
+      "loss": 0.1396,
+      "step": 2819
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.199136760330648e-05,
+      "loss": 0.1212,
+      "step": 2820
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1981622382237982e-05,
+      "loss": 0.11,
+      "step": 2821
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.19718784887673e-05,
+      "loss": 0.1279,
+      "step": 2822
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1962135927180181e-05,
+      "loss": 0.1298,
+      "step": 2823
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1952394701761772e-05,
+      "loss": 0.1155,
+      "step": 2824
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1942654816796643e-05,
+      "loss": 0.1224,
+      "step": 2825
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1932916276568752e-05,
+      "loss": 0.1366,
+      "step": 2826
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1923179085361503e-05,
+      "loss": 0.1234,
+      "step": 2827
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1913443247457665e-05,
+      "loss": 0.1074,
+      "step": 2828
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.190370876713944e-05,
+      "loss": 0.1173,
+      "step": 2829
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1893975648688426e-05,
+      "loss": 0.1009,
+      "step": 2830
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1884243896385622e-05,
+      "loss": 0.1445,
+      "step": 2831
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1874513514511414e-05,
+      "loss": 0.1048,
+      "step": 2832
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1864784507345605e-05,
+      "loss": 0.1513,
+      "step": 2833
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1855056879167382e-05,
+      "loss": 0.0991,
+      "step": 2834
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1845330634255327e-05,
+      "loss": 0.1035,
+      "step": 2835
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.183560577688742e-05,
+      "loss": 0.1465,
+      "step": 2836
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1825882311341013e-05,
+      "loss": 0.1072,
+      "step": 2837
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.181616024189287e-05,
+      "loss": 0.122,
+      "step": 2838
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1806439572819118e-05,
+      "loss": 0.1062,
+      "step": 2839
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1796720308395289e-05,
+      "loss": 0.0879,
+      "step": 2840
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1787002452896275e-05,
+      "loss": 0.1313,
+      "step": 2841
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1777286010596375e-05,
+      "loss": 0.1205,
+      "step": 2842
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.176757098576924e-05,
+      "loss": 0.1016,
+      "step": 2843
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1757857382687912e-05,
+      "loss": 0.1663,
+      "step": 2844
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1748145205624805e-05,
+      "loss": 0.1109,
+      "step": 2845
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.173843445885171e-05,
+      "loss": 0.092,
+      "step": 2846
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1728725146639775e-05,
+      "loss": 0.1112,
+      "step": 2847
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1719017273259538e-05,
+      "loss": 0.142,
+      "step": 2848
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1709310842980881e-05,
+      "loss": 0.1159,
+      "step": 2849
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1699605860073078e-05,
+      "loss": 0.1235,
+      "step": 2850
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1689902328804731e-05,
+      "loss": 0.1317,
+      "step": 2851
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1680200253443846e-05,
+      "loss": 0.1274,
+      "step": 2852
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1670499638257747e-05,
+      "loss": 0.0982,
+      "step": 2853
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.166080048751315e-05,
+      "loss": 0.0633,
+      "step": 2854
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1651102805476101e-05,
+      "loss": 0.1166,
+      "step": 2855
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1641406596412024e-05,
+      "loss": 0.1005,
+      "step": 2856
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1631711864585666e-05,
+      "loss": 0.147,
+      "step": 2857
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1622018614261155e-05,
+      "loss": 0.1246,
+      "step": 2858
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1612326849701943e-05,
+      "loss": 0.1577,
+      "step": 2859
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1602636575170848e-05,
+      "loss": 0.1828,
+      "step": 2860
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1592947794930012e-05,
+      "loss": 0.1345,
+      "step": 2861
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.158326051324094e-05,
+      "loss": 0.0939,
+      "step": 2862
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.157357473436446e-05,
+      "loss": 0.107,
+      "step": 2863
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1563890462560753e-05,
+      "loss": 0.0834,
+      "step": 2864
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1554207702089336e-05,
+      "loss": 0.1246,
+      "step": 2865
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1544526457209048e-05,
+      "loss": 0.1058,
+      "step": 2866
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1534846732178083e-05,
+      "loss": 0.1041,
+      "step": 2867
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1525168531253938e-05,
+      "loss": 0.0946,
+      "step": 2868
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.151549185869347e-05,
+      "loss": 0.119,
+      "step": 2869
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1505816718752838e-05,
+      "loss": 0.1,
+      "step": 2870
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1496143115687549e-05,
+      "loss": 0.0848,
+      "step": 2871
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1486471053752413e-05,
+      "loss": 0.1322,
+      "step": 2872
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1476800537201577e-05,
+      "loss": 0.1101,
+      "step": 2873
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1467131570288503e-05,
+      "loss": 0.1299,
+      "step": 2874
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.145746415726598e-05,
+      "loss": 0.1346,
+      "step": 2875
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1447798302386089e-05,
+      "loss": 0.1053,
+      "step": 2876
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.143813400990026e-05,
+      "loss": 0.112,
+      "step": 2877
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1428471284059205e-05,
+      "loss": 0.1539,
+      "step": 2878
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1418810129112968e-05,
+      "loss": 0.1099,
+      "step": 2879
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1409150549310889e-05,
+      "loss": 0.1417,
+      "step": 2880
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1399492548901629e-05,
+      "loss": 0.1045,
+      "step": 2881
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1389836132133132e-05,
+      "loss": 0.1163,
+      "step": 2882
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.138018130325267e-05,
+      "loss": 0.1423,
+      "step": 2883
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1370528066506798e-05,
+      "loss": 0.1058,
+      "step": 2884
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1360876426141392e-05,
+      "loss": 0.1166,
+      "step": 2885
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1351226386401596e-05,
+      "loss": 0.1466,
+      "step": 2886
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1341577951531875e-05,
+      "loss": 0.1264,
+      "step": 2887
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1331931125775974e-05,
+      "loss": 0.1129,
+      "step": 2888
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1322285913376946e-05,
+      "loss": 0.1211,
+      "step": 2889
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.131264231857711e-05,
+      "loss": 0.0992,
+      "step": 2890
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1303000345618103e-05,
+      "loss": 0.1208,
+      "step": 2891
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1293359998740816e-05,
+      "loss": 0.1135,
+      "step": 2892
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1283721282185453e-05,
+      "loss": 0.1296,
+      "step": 2893
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1274084200191491e-05,
+      "loss": 0.1238,
+      "step": 2894
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1264448756997678e-05,
+      "loss": 0.0942,
+      "step": 2895
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1254814956842066e-05,
+      "loss": 0.1347,
+      "step": 2896
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.124518280396195e-05,
+      "loss": 0.0889,
+      "step": 2897
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1235552302593931e-05,
+      "loss": 0.0993,
+      "step": 2898
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1225923456973863e-05,
+      "loss": 0.112,
+      "step": 2899
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1216296271336895e-05,
+      "loss": 0.0973,
+      "step": 2900
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.120667074991741e-05,
+      "loss": 0.099,
+      "step": 2901
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1197046896949097e-05,
+      "loss": 0.1283,
+      "step": 2902
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1187424716664885e-05,
+      "loss": 0.1283,
+      "step": 2903
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1177804213296985e-05,
+      "loss": 0.1137,
+      "step": 2904
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1168185391076853e-05,
+      "loss": 0.0865,
+      "step": 2905
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1158568254235224e-05,
+      "loss": 0.1134,
+      "step": 2906
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1148952807002072e-05,
+      "loss": 0.1498,
+      "step": 2907
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1139339053606647e-05,
+      "loss": 0.1339,
+      "step": 2908
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1129726998277435e-05,
+      "loss": 0.1194,
+      "step": 2909
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1120116645242202e-05,
+      "loss": 0.1142,
+      "step": 2910
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.111050799872793e-05,
+      "loss": 0.0955,
+      "step": 2911
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.110090106296088e-05,
+      "loss": 0.0845,
+      "step": 2912
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1091295842166544e-05,
+      "loss": 0.1394,
+      "step": 2913
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1081692340569674e-05,
+      "loss": 0.1301,
+      "step": 2914
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1072090562394242e-05,
+      "loss": 0.1192,
+      "step": 2915
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1062490511863495e-05,
+      "loss": 0.1255,
+      "step": 2916
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1052892193199884e-05,
+      "loss": 0.1005,
+      "step": 2917
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1043295610625133e-05,
+      "loss": 0.0878,
+      "step": 2918
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1033700768360172e-05,
+      "loss": 0.1042,
+      "step": 2919
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1024107670625193e-05,
+      "loss": 0.1441,
+      "step": 2920
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.101451632163959e-05,
+      "loss": 0.0987,
+      "step": 2921
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1004926725622021e-05,
+      "loss": 0.1252,
+      "step": 2922
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0995338886790345e-05,
+      "loss": 0.0851,
+      "step": 2923
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0985752809361665e-05,
+      "loss": 0.1042,
+      "step": 2924
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0976168497552307e-05,
+      "loss": 0.1412,
+      "step": 2925
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0966585955577814e-05,
+      "loss": 0.1109,
+      "step": 2926
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0957005187652954e-05,
+      "loss": 0.0846,
+      "step": 2927
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0947426197991714e-05,
+      "loss": 0.1099,
+      "step": 2928
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0937848990807303e-05,
+      "loss": 0.1402,
+      "step": 2929
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.092827357031214e-05,
+      "loss": 0.0775,
+      "step": 2930
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0918699940717866e-05,
+      "loss": 0.1288,
+      "step": 2931
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.090912810623532e-05,
+      "loss": 0.1158,
+      "step": 2932
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0899558071074568e-05,
+      "loss": 0.1215,
+      "step": 2933
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0889989839444873e-05,
+      "loss": 0.118,
+      "step": 2934
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0880423415554716e-05,
+      "loss": 0.1126,
+      "step": 2935
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0870858803611765e-05,
+      "loss": 0.1191,
+      "step": 2936
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0861296007822908e-05,
+      "loss": 0.1066,
+      "step": 2937
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0851735032394227e-05,
+      "loss": 0.1097,
+      "step": 2938
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0842175881531008e-05,
+      "loss": 0.1049,
+      "step": 2939
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.083261855943772e-05,
+      "loss": 0.1565,
+      "step": 2940
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0823063070318055e-05,
+      "loss": 0.0793,
+      "step": 2941
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0813509418374869e-05,
+      "loss": 0.1154,
+      "step": 2942
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.080395760781023e-05,
+      "loss": 0.1202,
+      "step": 2943
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0794407642825383e-05,
+      "loss": 0.1418,
+      "step": 2944
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0784859527620779e-05,
+      "loss": 0.0972,
+      "step": 2945
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0775313266396028e-05,
+      "loss": 0.1357,
+      "step": 2946
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0765768863349955e-05,
+      "loss": 0.1273,
+      "step": 2947
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0756226322680542e-05,
+      "loss": 0.1306,
+      "step": 2948
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0746685648584973e-05,
+      "loss": 0.1607,
+      "step": 2949
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0737146845259592e-05,
+      "loss": 0.1094,
+      "step": 2950
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0727609916899936e-05,
+      "loss": 0.1114,
+      "step": 2951
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0718074867700704e-05,
+      "loss": 0.1101,
+      "step": 2952
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.070854170185578e-05,
+      "loss": 0.1101,
+      "step": 2953
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0699010423558218e-05,
+      "loss": 0.1063,
+      "step": 2954
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0689481037000228e-05,
+      "loss": 0.1059,
+      "step": 2955
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0679953546373213e-05,
+      "loss": 0.146,
+      "step": 2956
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.067042795586771e-05,
+      "loss": 0.1526,
+      "step": 2957
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0660904269673455e-05,
+      "loss": 0.0854,
+      "step": 2958
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0651382491979313e-05,
+      "loss": 0.1058,
+      "step": 2959
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0641862626973342e-05,
+      "loss": 0.1242,
+      "step": 2960
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0632344678842726e-05,
+      "loss": 0.0838,
+      "step": 2961
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0622828651773834e-05,
+      "loss": 0.099,
+      "step": 2962
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0613314549952172e-05,
+      "loss": 0.1176,
+      "step": 2963
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0603802377562416e-05,
+      "loss": 0.1334,
+      "step": 2964
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0594292138788367e-05,
+      "loss": 0.1489,
+      "step": 2965
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0584783837813001e-05,
+      "loss": 0.1152,
+      "step": 2966
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.057527747881843e-05,
+      "loss": 0.0982,
+      "step": 2967
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0565773065985919e-05,
+      "loss": 0.1246,
+      "step": 2968
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0556270603495861e-05,
+      "loss": 0.1226,
+      "step": 2969
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0546770095527817e-05,
+      "loss": 0.101,
+      "step": 2970
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0537271546260457e-05,
+      "loss": 0.1297,
+      "step": 2971
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.052777495987162e-05,
+      "loss": 0.1485,
+      "step": 2972
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0518280340538257e-05,
+      "loss": 0.0974,
+      "step": 2973
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0508787692436476e-05,
+      "loss": 0.0901,
+      "step": 2974
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0499297019741495e-05,
+      "loss": 0.1036,
+      "step": 2975
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0489808326627681e-05,
+      "loss": 0.1384,
+      "step": 2976
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0480321617268517e-05,
+      "loss": 0.1139,
+      "step": 2977
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0470836895836634e-05,
+      "loss": 0.1151,
+      "step": 2978
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0461354166503755e-05,
+      "loss": 0.119,
+      "step": 2979
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.045187343344077e-05,
+      "loss": 0.1092,
+      "step": 2980
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0442394700817646e-05,
+      "loss": 0.0697,
+      "step": 2981
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0432917972803506e-05,
+      "loss": 0.1416,
+      "step": 2982
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0423443253566569e-05,
+      "loss": 0.1265,
+      "step": 2983
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0413970547274184e-05,
+      "loss": 0.1009,
+      "step": 2984
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0404499858092815e-05,
+      "loss": 0.0843,
+      "step": 2985
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0395031190188015e-05,
+      "loss": 0.0949,
+      "step": 2986
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0385564547724481e-05,
+      "loss": 0.1306,
+      "step": 2987
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0376099934866e-05,
+      "loss": 0.1139,
+      "step": 2988
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.036663735577547e-05,
+      "loss": 0.1388,
+      "step": 2989
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0357176814614895e-05,
+      "loss": 0.0696,
+      "step": 2990
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0347718315545382e-05,
+      "loss": 0.0836,
+      "step": 2991
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0338261862727139e-05,
+      "loss": 0.105,
+      "step": 2992
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.032880746031948e-05,
+      "loss": 0.1088,
+      "step": 2993
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0319355112480805e-05,
+      "loss": 0.1189,
+      "step": 2994
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.030990482336863e-05,
+      "loss": 0.1052,
+      "step": 2995
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.030045659713954e-05,
+      "loss": 0.138,
+      "step": 2996
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0291010437949235e-05,
+      "loss": 0.1469,
+      "step": 2997
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0281566349952489e-05,
+      "loss": 0.1506,
+      "step": 2998
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0272124337303185e-05,
+      "loss": 0.1381,
+      "step": 2999
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0262684404154265e-05,
+      "loss": 0.0918,
+      "step": 3000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0253246554657785e-05,
+      "loss": 0.0963,
+      "step": 3001
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0243810792964865e-05,
+      "loss": 0.1174,
+      "step": 3002
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0234377123225721e-05,
+      "loss": 0.1222,
+      "step": 3003
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.022494554958963e-05,
+      "loss": 0.0762,
+      "step": 3004
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0215516076204972e-05,
+      "loss": 0.1208,
+      "step": 3005
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0206088707219176e-05,
+      "loss": 0.1047,
+      "step": 3006
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0196663446778769e-05,
+      "loss": 0.1098,
+      "step": 3007
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0187240299029336e-05,
+      "loss": 0.1167,
+      "step": 3008
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0177819268115543e-05,
+      "loss": 0.143,
+      "step": 3009
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0168400358181113e-05,
+      "loss": 0.0927,
+      "step": 3010
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0158983573368847e-05,
+      "loss": 0.131,
+      "step": 3011
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0149568917820601e-05,
+      "loss": 0.1188,
+      "step": 3012
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.014015639567731e-05,
+      "loss": 0.1121,
+      "step": 3013
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.013074601107895e-05,
+      "loss": 0.0981,
+      "step": 3014
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0121337768164573e-05,
+      "loss": 0.1404,
+      "step": 3015
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0111931671072287e-05,
+      "loss": 0.1559,
+      "step": 3016
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0102527723939244e-05,
+      "loss": 0.1275,
+      "step": 3017
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.009312593090167e-05,
+      "loss": 0.1024,
+      "step": 3018
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0083726296094825e-05,
+      "loss": 0.1292,
+      "step": 3019
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0074328823653038e-05,
+      "loss": 0.1124,
+      "step": 3020
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.006493351770966e-05,
+      "loss": 0.1095,
+      "step": 3021
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.005554038239712e-05,
+      "loss": 0.1035,
+      "step": 3022
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.004614942184687e-05,
+      "loss": 0.0871,
+      "step": 3023
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0036760640189418e-05,
+      "loss": 0.0839,
+      "step": 3024
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0027374041554304e-05,
+      "loss": 0.116,
+      "step": 3025
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0017989630070116e-05,
+      "loss": 0.1026,
+      "step": 3026
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0008607409864473e-05,
+      "loss": 0.106,
+      "step": 3027
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.999227385064045e-06,
+      "loss": 0.1754,
+      "step": 3028
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.989849559794513e-06,
+      "loss": 0.1097,
+      "step": 3029
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.980473938180607e-06,
+      "loss": 0.0993,
+      "step": 3030
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.971100524346085e-06,
+      "loss": 0.139,
+      "step": 3031
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.961729322413733e-06,
+      "loss": 0.1332,
+      "step": 3032
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.952360336505358e-06,
+      "loss": 0.0979,
+      "step": 3033
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.942993570741812e-06,
+      "loss": 0.1466,
+      "step": 3034
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.933629029242939e-06,
+      "loss": 0.1116,
+      "step": 3035
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.924266716127633e-06,
+      "loss": 0.1174,
+      "step": 3036
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.914906635513793e-06,
+      "loss": 0.1287,
+      "step": 3037
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.905548791518347e-06,
+      "loss": 0.0876,
+      "step": 3038
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.89619318825722e-06,
+      "loss": 0.1326,
+      "step": 3039
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.886839829845375e-06,
+      "loss": 0.1081,
+      "step": 3040
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.87748872039677e-06,
+      "loss": 0.0791,
+      "step": 3041
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.868139864024389e-06,
+      "loss": 0.1045,
+      "step": 3042
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.858793264840201e-06,
+      "loss": 0.1102,
+      "step": 3043
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.849448926955204e-06,
+      "loss": 0.1373,
+      "step": 3044
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.840106854479408e-06,
+      "loss": 0.1201,
+      "step": 3045
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.830767051521796e-06,
+      "loss": 0.14,
+      "step": 3046
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.821429522190379e-06,
+      "loss": 0.141,
+      "step": 3047
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.812094270592155e-06,
+      "loss": 0.1044,
+      "step": 3048
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.802761300833133e-06,
+      "loss": 0.1232,
+      "step": 3049
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.793430617018298e-06,
+      "loss": 0.1312,
+      "step": 3050
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.78410222325165e-06,
+      "loss": 0.1488,
+      "step": 3051
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.774776123636166e-06,
+      "loss": 0.119,
+      "step": 3052
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.765452322273831e-06,
+      "loss": 0.1085,
+      "step": 3053
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.756130823265596e-06,
+      "loss": 0.0957,
+      "step": 3054
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.746811630711423e-06,
+      "loss": 0.087,
+      "step": 3055
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.737494748710243e-06,
+      "loss": 0.1362,
+      "step": 3056
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.728180181359983e-06,
+      "loss": 0.1145,
+      "step": 3057
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.718867932757538e-06,
+      "loss": 0.1284,
+      "step": 3058
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.709558006998799e-06,
+      "loss": 0.0734,
+      "step": 3059
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.700250408178619e-06,
+      "loss": 0.0774,
+      "step": 3060
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.690945140390841e-06,
+      "loss": 0.1413,
+      "step": 3061
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.681642207728275e-06,
+      "loss": 0.1477,
+      "step": 3062
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.672341614282713e-06,
+      "loss": 0.1169,
+      "step": 3063
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.663043364144897e-06,
+      "loss": 0.1531,
+      "step": 3064
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.653747461404566e-06,
+      "loss": 0.1093,
+      "step": 3065
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.644453910150406e-06,
+      "loss": 0.1195,
+      "step": 3066
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.635162714470082e-06,
+      "loss": 0.1063,
+      "step": 3067
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.62587387845021e-06,
+      "loss": 0.1287,
+      "step": 3068
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.616587406176383e-06,
+      "loss": 0.1106,
+      "step": 3069
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.607303301733134e-06,
+      "loss": 0.1297,
+      "step": 3070
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.598021569203982e-06,
+      "loss": 0.088,
+      "step": 3071
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.588742212671376e-06,
+      "loss": 0.1547,
+      "step": 3072
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.579465236216743e-06,
+      "loss": 0.0524,
+      "step": 3073
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.57019064392044e-06,
+      "loss": 0.1053,
+      "step": 3074
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.56091843986179e-06,
+      "loss": 0.1175,
+      "step": 3075
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.551648628119066e-06,
+      "loss": 0.1241,
+      "step": 3076
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.542381212769484e-06,
+      "loss": 0.1066,
+      "step": 3077
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.533116197889213e-06,
+      "loss": 0.1349,
+      "step": 3078
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.52385358755335e-06,
+      "loss": 0.0878,
+      "step": 3079
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.514593385835952e-06,
+      "loss": 0.1091,
+      "step": 3080
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.505335596810005e-06,
+      "loss": 0.1065,
+      "step": 3081
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.496080224547449e-06,
+      "loss": 0.1148,
+      "step": 3082
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.486827273119129e-06,
+      "loss": 0.139,
+      "step": 3083
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.477576746594875e-06,
+      "loss": 0.1025,
+      "step": 3084
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.468328649043397e-06,
+      "loss": 0.0981,
+      "step": 3085
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.459082984532376e-06,
+      "loss": 0.0936,
+      "step": 3086
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.449839757128403e-06,
+      "loss": 0.0971,
+      "step": 3087
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.440598970897011e-06,
+      "loss": 0.0709,
+      "step": 3088
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.431360629902634e-06,
+      "loss": 0.1215,
+      "step": 3089
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.422124738208665e-06,
+      "loss": 0.1019,
+      "step": 3090
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.41289129987739e-06,
+      "loss": 0.1003,
+      "step": 3091
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.40366031897004e-06,
+      "loss": 0.1245,
+      "step": 3092
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.394431799546736e-06,
+      "loss": 0.1068,
+      "step": 3093
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.38520574566655e-06,
+      "loss": 0.1081,
+      "step": 3094
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.375982161387443e-06,
+      "loss": 0.1003,
+      "step": 3095
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.366761050766305e-06,
+      "loss": 0.0784,
+      "step": 3096
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.35754241785893e-06,
+      "loss": 0.1084,
+      "step": 3097
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.34832626672003e-06,
+      "loss": 0.1317,
+      "step": 3098
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.339112601403213e-06,
+      "loss": 0.0839,
+      "step": 3099
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.329901425961007e-06,
+      "loss": 0.1013,
+      "step": 3100
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.320692744444835e-06,
+      "loss": 0.1458,
+      "step": 3101
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.311486560905032e-06,
+      "loss": 0.1112,
+      "step": 3102
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.30228287939082e-06,
+      "loss": 0.0951,
+      "step": 3103
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.293081703950338e-06,
+      "loss": 0.1284,
+      "step": 3104
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.283883038630616e-06,
+      "loss": 0.1602,
+      "step": 3105
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.27468688747757e-06,
+      "loss": 0.0861,
+      "step": 3106
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.265493254536032e-06,
+      "loss": 0.1025,
+      "step": 3107
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.256302143849693e-06,
+      "loss": 0.1122,
+      "step": 3108
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.24711355946118e-06,
+      "loss": 0.1299,
+      "step": 3109
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.23792750541196e-06,
+      "loss": 0.1676,
+      "step": 3110
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.228743985742425e-06,
+      "loss": 0.1203,
+      "step": 3111
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.21956300449183e-06,
+      "loss": 0.1532,
+      "step": 3112
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.210384565698334e-06,
+      "loss": 0.1639,
+      "step": 3113
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.201208673398948e-06,
+      "loss": 0.0838,
+      "step": 3114
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.192035331629593e-06,
+      "loss": 0.1054,
+      "step": 3115
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.182864544425044e-06,
+      "loss": 0.0914,
+      "step": 3116
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.17369631581898e-06,
+      "loss": 0.1071,
+      "step": 3117
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.16453064984392e-06,
+      "loss": 0.093,
+      "step": 3118
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.155367550531289e-06,
+      "loss": 0.1055,
+      "step": 3119
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.146207021911356e-06,
+      "loss": 0.1266,
+      "step": 3120
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.137049068013284e-06,
+      "loss": 0.1218,
+      "step": 3121
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.127893692865075e-06,
+      "loss": 0.1188,
+      "step": 3122
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.118740900493634e-06,
+      "loss": 0.1213,
+      "step": 3123
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.10959069492469e-06,
+      "loss": 0.1095,
+      "step": 3124
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.100443080182862e-06,
+      "loss": 0.0975,
+      "step": 3125
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.091298060291617e-06,
+      "loss": 0.1213,
+      "step": 3126
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.082155639273293e-06,
+      "loss": 0.0938,
+      "step": 3127
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.073015821149064e-06,
+      "loss": 0.0959,
+      "step": 3128
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.063878609938979e-06,
+      "loss": 0.1237,
+      "step": 3129
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.054744009661926e-06,
+      "loss": 0.134,
+      "step": 3130
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.045612024335661e-06,
+      "loss": 0.0848,
+      "step": 3131
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.03648265797677e-06,
+      "loss": 0.1112,
+      "step": 3132
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.027355914600707e-06,
+      "loss": 0.0782,
+      "step": 3133
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.018231798221751e-06,
+      "loss": 0.0878,
+      "step": 3134
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.009110312853046e-06,
+      "loss": 0.1057,
+      "step": 3135
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.99999146250657e-06,
+      "loss": 0.1157,
+      "step": 3136
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.990875251193135e-06,
+      "loss": 0.1386,
+      "step": 3137
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.981761682922409e-06,
+      "loss": 0.1732,
+      "step": 3138
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.972650761702876e-06,
+      "loss": 0.1006,
+      "step": 3139
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.963542491541877e-06,
+      "loss": 0.0781,
+      "step": 3140
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.954436876445572e-06,
+      "loss": 0.0931,
+      "step": 3141
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.945333920418963e-06,
+      "loss": 0.1125,
+      "step": 3142
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.936233627465873e-06,
+      "loss": 0.1222,
+      "step": 3143
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.927136001588959e-06,
+      "loss": 0.0999,
+      "step": 3144
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.918041046789704e-06,
+      "loss": 0.0937,
+      "step": 3145
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.908948767068427e-06,
+      "loss": 0.1079,
+      "step": 3146
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.899859166424241e-06,
+      "loss": 0.081,
+      "step": 3147
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.89077224885512e-06,
+      "loss": 0.1094,
+      "step": 3148
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.881688018357824e-06,
+      "loss": 0.0671,
+      "step": 3149
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.872606478927951e-06,
+      "loss": 0.0918,
+      "step": 3150
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.863527634559902e-06,
+      "loss": 0.1155,
+      "step": 3151
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.854451489246912e-06,
+      "loss": 0.1327,
+      "step": 3152
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.845378046981002e-06,
+      "loss": 0.1453,
+      "step": 3153
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.836307311753032e-06,
+      "loss": 0.1297,
+      "step": 3154
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.827239287552646e-06,
+      "loss": 0.1429,
+      "step": 3155
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.81817397836832e-06,
+      "loss": 0.1272,
+      "step": 3156
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.80911138818731e-06,
+      "loss": 0.111,
+      "step": 3157
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.800051520995703e-06,
+      "loss": 0.116,
+      "step": 3158
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.790994380778364e-06,
+      "loss": 0.115,
+      "step": 3159
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.781939971518974e-06,
+      "loss": 0.1276,
+      "step": 3160
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.772888297200006e-06,
+      "loss": 0.1031,
+      "step": 3161
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.76383936180274e-06,
+      "loss": 0.1112,
+      "step": 3162
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.754793169307233e-06,
+      "loss": 0.0801,
+      "step": 3163
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.74574972369235e-06,
+      "loss": 0.0937,
+      "step": 3164
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.736709028935748e-06,
+      "loss": 0.1192,
+      "step": 3165
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.72767108901386e-06,
+      "loss": 0.1069,
+      "step": 3166
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.718635907901931e-06,
+      "loss": 0.1394,
+      "step": 3167
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.709603489573973e-06,
+      "loss": 0.0931,
+      "step": 3168
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.700573838002787e-06,
+      "loss": 0.0949,
+      "step": 3169
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.691546957159958e-06,
+      "loss": 0.1293,
+      "step": 3170
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.682522851015856e-06,
+      "loss": 0.1007,
+      "step": 3171
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.673501523539622e-06,
+      "loss": 0.1689,
+      "step": 3172
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.66448297869919e-06,
+      "loss": 0.1555,
+      "step": 3173
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.655467220461255e-06,
+      "loss": 0.0885,
+      "step": 3174
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.646454252791293e-06,
+      "loss": 0.0904,
+      "step": 3175
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.637444079653552e-06,
+      "loss": 0.0993,
+      "step": 3176
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.628436705011049e-06,
+      "loss": 0.0897,
+      "step": 3177
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.619432132825568e-06,
+      "loss": 0.1129,
+      "step": 3178
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.610430367057679e-06,
+      "loss": 0.141,
+      "step": 3179
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.601431411666683e-06,
+      "loss": 0.0716,
+      "step": 3180
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.592435270610676e-06,
+      "loss": 0.1311,
+      "step": 3181
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.583441947846502e-06,
+      "loss": 0.0791,
+      "step": 3182
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.574451447329772e-06,
+      "loss": 0.067,
+      "step": 3183
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.565463773014844e-06,
+      "loss": 0.0915,
+      "step": 3184
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.556478928854846e-06,
+      "loss": 0.1207,
+      "step": 3185
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.54749691880165e-06,
+      "loss": 0.1229,
+      "step": 3186
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.538517746805896e-06,
+      "loss": 0.1185,
+      "step": 3187
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.529541416816962e-06,
+      "loss": 0.0787,
+      "step": 3188
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.520567932782984e-06,
+      "loss": 0.1676,
+      "step": 3189
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.511597298650839e-06,
+      "loss": 0.1342,
+      "step": 3190
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.502629518366156e-06,
+      "loss": 0.1141,
+      "step": 3191
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.493664595873302e-06,
+      "loss": 0.095,
+      "step": 3192
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.484702535115411e-06,
+      "loss": 0.1092,
+      "step": 3193
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.475743340034314e-06,
+      "loss": 0.106,
+      "step": 3194
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.466787014570628e-06,
+      "loss": 0.1004,
+      "step": 3195
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.457833562663678e-06,
+      "loss": 0.1246,
+      "step": 3196
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.448882988251531e-06,
+      "loss": 0.0915,
+      "step": 3197
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.439935295271006e-06,
+      "loss": 0.1011,
+      "step": 3198
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.430990487657627e-06,
+      "loss": 0.1368,
+      "step": 3199
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.422048569345667e-06,
+      "loss": 0.0985,
+      "step": 3200
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.413109544268124e-06,
+      "loss": 0.0953,
+      "step": 3201
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.404173416356722e-06,
+      "loss": 0.1204,
+      "step": 3202
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.395240189541904e-06,
+      "loss": 0.078,
+      "step": 3203
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.386309867752863e-06,
+      "loss": 0.1156,
+      "step": 3204
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.377382454917472e-06,
+      "loss": 0.0829,
+      "step": 3205
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.36845795496237e-06,
+      "loss": 0.092,
+      "step": 3206
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.35953637181288e-06,
+      "loss": 0.1192,
+      "step": 3207
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.350617709393062e-06,
+      "loss": 0.101,
+      "step": 3208
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.34170197162568e-06,
+      "loss": 0.1144,
+      "step": 3209
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.332789162432218e-06,
+      "loss": 0.1187,
+      "step": 3210
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.323879285732864e-06,
+      "loss": 0.1062,
+      "step": 3211
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.314972345446535e-06,
+      "loss": 0.1332,
+      "step": 3212
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.306068345490835e-06,
+      "loss": 0.0819,
+      "step": 3213
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.297167289782088e-06,
+      "loss": 0.1134,
+      "step": 3214
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.288269182235314e-06,
+      "loss": 0.1072,
+      "step": 3215
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.279374026764242e-06,
+      "loss": 0.1326,
+      "step": 3216
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.270481827281297e-06,
+      "loss": 0.132,
+      "step": 3217
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.261592587697626e-06,
+      "loss": 0.1394,
+      "step": 3218
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.25270631192303e-06,
+      "loss": 0.1346,
+      "step": 3219
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.243823003866052e-06,
+      "loss": 0.1354,
+      "step": 3220
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.234942667433903e-06,
+      "loss": 0.1005,
+      "step": 3221
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.226065306532497e-06,
+      "loss": 0.0907,
+      "step": 3222
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.217190925066434e-06,
+      "loss": 0.1266,
+      "step": 3223
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.20831952693901e-06,
+      "loss": 0.0865,
+      "step": 3224
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.199451116052193e-06,
+      "loss": 0.07,
+      "step": 3225
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.190585696306666e-06,
+      "loss": 0.1564,
+      "step": 3226
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.181723271601771e-06,
+      "loss": 0.0958,
+      "step": 3227
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.172863845835537e-06,
+      "loss": 0.1148,
+      "step": 3228
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.164007422904695e-06,
+      "loss": 0.1009,
+      "step": 3229
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.15515400670461e-06,
+      "loss": 0.1323,
+      "step": 3230
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.14630360112938e-06,
+      "loss": 0.1117,
+      "step": 3231
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.137456210071736e-06,
+      "loss": 0.0952,
+      "step": 3232
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.128611837423105e-06,
+      "loss": 0.0879,
+      "step": 3233
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.119770487073578e-06,
+      "loss": 0.1178,
+      "step": 3234
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.11093216291192e-06,
+      "loss": 0.1332,
+      "step": 3235
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.102096868825556e-06,
+      "loss": 0.1031,
+      "step": 3236
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.093264608700596e-06,
+      "loss": 0.0989,
+      "step": 3237
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.084435386421805e-06,
+      "loss": 0.0834,
+      "step": 3238
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.075609205872609e-06,
+      "loss": 0.1116,
+      "step": 3239
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.066786070935098e-06,
+      "loss": 0.1109,
+      "step": 3240
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.057965985490028e-06,
+      "loss": 0.0789,
+      "step": 3241
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.0491489534168e-06,
+      "loss": 0.1616,
+      "step": 3242
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.040334978593501e-06,
+      "loss": 0.101,
+      "step": 3243
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.03152406489683e-06,
+      "loss": 0.1369,
+      "step": 3244
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.022716216202184e-06,
+      "loss": 0.0804,
+      "step": 3245
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.013911436383581e-06,
+      "loss": 0.0969,
+      "step": 3246
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.005109729313703e-06,
+      "loss": 0.1151,
+      "step": 3247
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.996311098863875e-06,
+      "loss": 0.08,
+      "step": 3248
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.98751554890407e-06,
+      "loss": 0.1366,
+      "step": 3249
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.978723083302909e-06,
+      "loss": 0.1138,
+      "step": 3250
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.969933705927653e-06,
+      "loss": 0.1157,
+      "step": 3251
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.961147420644211e-06,
+      "loss": 0.1162,
+      "step": 3252
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.952364231317122e-06,
+      "loss": 0.1117,
+      "step": 3253
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.94358414180957e-06,
+      "loss": 0.137,
+      "step": 3254
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.934807155983367e-06,
+      "loss": 0.1487,
+      "step": 3255
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.92603327769898e-06,
+      "loss": 0.0991,
+      "step": 3256
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.91726251081549e-06,
+      "loss": 0.124,
+      "step": 3257
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.908494859190614e-06,
+      "loss": 0.0892,
+      "step": 3258
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.8997303266807e-06,
+      "loss": 0.1336,
+      "step": 3259
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.890968917140724e-06,
+      "loss": 0.0788,
+      "step": 3260
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.882210634424287e-06,
+      "loss": 0.1026,
+      "step": 3261
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.873455482383629e-06,
+      "loss": 0.1295,
+      "step": 3262
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.864703464869579e-06,
+      "loss": 0.0832,
+      "step": 3263
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.855954585731628e-06,
+      "loss": 0.1187,
+      "step": 3264
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.84720884881786e-06,
+      "loss": 0.1148,
+      "step": 3265
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.838466257974984e-06,
+      "loss": 0.1506,
+      "step": 3266
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.82972681704832e-06,
+      "loss": 0.1229,
+      "step": 3267
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.820990529881826e-06,
+      "loss": 0.1132,
+      "step": 3268
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.812257400318029e-06,
+      "loss": 0.0643,
+      "step": 3269
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.803527432198115e-06,
+      "loss": 0.1243,
+      "step": 3270
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.794800629361848e-06,
+      "loss": 0.0845,
+      "step": 3271
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.786076995647613e-06,
+      "loss": 0.1334,
+      "step": 3272
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.777356534892394e-06,
+      "loss": 0.104,
+      "step": 3273
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.768639250931787e-06,
+      "loss": 0.1293,
+      "step": 3274
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.759925147599974e-06,
+      "loss": 0.1208,
+      "step": 3275
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.751214228729768e-06,
+      "loss": 0.1065,
+      "step": 3276
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.742506498152557e-06,
+      "loss": 0.1186,
+      "step": 3277
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.733801959698334e-06,
+      "loss": 0.1051,
+      "step": 3278
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.725100617195687e-06,
+      "loss": 0.1136,
+      "step": 3279
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.716402474471799e-06,
+      "loss": 0.1929,
+      "step": 3280
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.707707535352438e-06,
+      "loss": 0.1063,
+      "step": 3281
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.69901580366199e-06,
+      "loss": 0.0741,
+      "step": 3282
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.690327283223388e-06,
+      "loss": 0.0734,
+      "step": 3283
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.681641977858194e-06,
+      "loss": 0.0558,
+      "step": 3284
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.672959891386525e-06,
+      "loss": 0.1172,
+      "step": 3285
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.664281027627094e-06,
+      "loss": 0.0825,
+      "step": 3286
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.655605390397216e-06,
+      "loss": 0.0855,
+      "step": 3287
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.646932983512739e-06,
+      "loss": 0.1124,
+      "step": 3288
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.63826381078814e-06,
+      "loss": 0.1024,
+      "step": 3289
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.629597876036447e-06,
+      "loss": 0.0756,
+      "step": 3290
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.620935183069267e-06,
+      "loss": 0.132,
+      "step": 3291
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.61227573569678e-06,
+      "loss": 0.1279,
+      "step": 3292
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.603619537727755e-06,
+      "loss": 0.0991,
+      "step": 3293
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.594966592969501e-06,
+      "loss": 0.0836,
+      "step": 3294
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.5863169052279284e-06,
+      "loss": 0.1078,
+      "step": 3295
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.577670478307494e-06,
+      "loss": 0.0824,
+      "step": 3296
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.5690273160112285e-06,
+      "loss": 0.1227,
+      "step": 3297
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.5603874221407235e-06,
+      "loss": 0.0969,
+      "step": 3298
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.551750800496136e-06,
+      "loss": 0.1334,
+      "step": 3299
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.543117454876174e-06,
+      "loss": 0.1323,
+      "step": 3300
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.534487389078133e-06,
+      "loss": 0.1158,
+      "step": 3301
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.5258606068978234e-06,
+      "loss": 0.1177,
+      "step": 3302
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.517237112129646e-06,
+      "loss": 0.0793,
+      "step": 3303
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.50861690856654e-06,
+      "loss": 0.1217,
+      "step": 3304
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.500000000000004e-06,
+      "loss": 0.1169,
+      "step": 3305
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.491386390220073e-06,
+      "loss": 0.1153,
+      "step": 3306
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.48277608301536e-06,
+      "loss": 0.1078,
+      "step": 3307
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.474169082172986e-06,
+      "loss": 0.0942,
+      "step": 3308
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.465565391478656e-06,
+      "loss": 0.1146,
+      "step": 3309
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.456965014716594e-06,
+      "loss": 0.1093,
+      "step": 3310
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.448367955669576e-06,
+      "loss": 0.1148,
+      "step": 3311
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.439774218118917e-06,
+      "loss": 0.1089,
+      "step": 3312
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.431183805844471e-06,
+      "loss": 0.0915,
+      "step": 3313
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.422596722624623e-06,
+      "loss": 0.1027,
+      "step": 3314
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.414012972236315e-06,
+      "loss": 0.1035,
+      "step": 3315
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.405432558454999e-06,
+      "loss": 0.0842,
+      "step": 3316
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.396855485054665e-06,
+      "loss": 0.1417,
+      "step": 3317
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.388281755807853e-06,
+      "loss": 0.112,
+      "step": 3318
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.379711374485596e-06,
+      "loss": 0.1156,
+      "step": 3319
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.3711443448574925e-06,
+      "loss": 0.0981,
+      "step": 3320
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.362580670691642e-06,
+      "loss": 0.0927,
+      "step": 3321
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.354020355754676e-06,
+      "loss": 0.127,
+      "step": 3322
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.34546340381175e-06,
+      "loss": 0.1059,
+      "step": 3323
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.336909818626537e-06,
+      "loss": 0.1098,
+      "step": 3324
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.328359603961226e-06,
+      "loss": 0.0803,
+      "step": 3325
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.319812763576543e-06,
+      "loss": 0.1345,
+      "step": 3326
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.311269301231694e-06,
+      "loss": 0.1018,
+      "step": 3327
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.3027292206844374e-06,
+      "loss": 0.1331,
+      "step": 3328
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.294192525691021e-06,
+      "loss": 0.1173,
+      "step": 3329
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.285659220006208e-06,
+      "loss": 0.1074,
+      "step": 3330
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.277129307383269e-06,
+      "loss": 0.1314,
+      "step": 3331
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.268602791574002e-06,
+      "loss": 0.1153,
+      "step": 3332
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.26007967632867e-06,
+      "loss": 0.1067,
+      "step": 3333
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.251559965396083e-06,
+      "loss": 0.0817,
+      "step": 3334
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.243043662523532e-06,
+      "loss": 0.0763,
+      "step": 3335
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.234530771456809e-06,
+      "loss": 0.1232,
+      "step": 3336
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.22602129594021e-06,
+      "loss": 0.1225,
+      "step": 3337
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.217515239716529e-06,
+      "loss": 0.0965,
+      "step": 3338
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.209012606527046e-06,
+      "loss": 0.0674,
+      "step": 3339
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.200513400111564e-06,
+      "loss": 0.1086,
+      "step": 3340
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.192017624208335e-06,
+      "loss": 0.1309,
+      "step": 3341
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.183525282554142e-06,
+      "loss": 0.1001,
+      "step": 3342
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.175036378884236e-06,
+      "loss": 0.1194,
+      "step": 3343
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.166550916932363e-06,
+      "loss": 0.1077,
+      "step": 3344
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.1580689004307454e-06,
+      "loss": 0.1076,
+      "step": 3345
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.149590333110109e-06,
+      "loss": 0.1086,
+      "step": 3346
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.141115218699651e-06,
+      "loss": 0.1125,
+      "step": 3347
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.132643560927045e-06,
+      "loss": 0.1034,
+      "step": 3348
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.124175363518453e-06,
+      "loss": 0.1147,
+      "step": 3349
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.115710630198504e-06,
+      "loss": 0.1123,
+      "step": 3350
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.10724936469033e-06,
+      "loss": 0.1181,
+      "step": 3351
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.098791570715496e-06,
+      "loss": 0.1158,
+      "step": 3352
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.09033725199408e-06,
+      "loss": 0.0863,
+      "step": 3353
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.081886412244608e-06,
+      "loss": 0.1266,
+      "step": 3354
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.073439055184081e-06,
+      "loss": 0.0945,
+      "step": 3355
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.064995184527969e-06,
+      "loss": 0.1079,
+      "step": 3356
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.05655480399022e-06,
+      "loss": 0.071,
+      "step": 3357
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.0481179172832165e-06,
+      "loss": 0.1104,
+      "step": 3358
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.039684528117839e-06,
+      "loss": 0.0512,
+      "step": 3359
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.0312546402034104e-06,
+      "loss": 0.0842,
+      "step": 3360
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.022828257247717e-06,
+      "loss": 0.1075,
+      "step": 3361
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.014405382957003e-06,
+      "loss": 0.1184,
+      "step": 3362
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.005986021035971e-06,
+      "loss": 0.1089,
+      "step": 3363
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.997570175187775e-06,
+      "loss": 0.1218,
+      "step": 3364
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.9891578491140376e-06,
+      "loss": 0.1125,
+      "step": 3365
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.980749046514801e-06,
+      "loss": 0.1347,
+      "step": 3366
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.972343771088598e-06,
+      "loss": 0.0867,
+      "step": 3367
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.96394202653238e-06,
+      "loss": 0.0716,
+      "step": 3368
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.955543816541555e-06,
+      "loss": 0.1316,
+      "step": 3369
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.947149144809975e-06,
+      "loss": 0.1123,
+      "step": 3370
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.938758015029952e-06,
+      "loss": 0.0873,
+      "step": 3371
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.930370430892203e-06,
+      "loss": 0.0928,
+      "step": 3372
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.921986396085926e-06,
+      "loss": 0.1036,
+      "step": 3373
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.913605914298731e-06,
+      "loss": 0.1257,
+      "step": 3374
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.905228989216677e-06,
+      "loss": 0.1328,
+      "step": 3375
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.896855624524257e-06,
+      "loss": 0.1192,
+      "step": 3376
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.888485823904388e-06,
+      "loss": 0.0936,
+      "step": 3377
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.880119591038443e-06,
+      "loss": 0.0858,
+      "step": 3378
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.871756929606202e-06,
+      "loss": 0.1608,
+      "step": 3379
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.863397843285884e-06,
+      "loss": 0.0999,
+      "step": 3380
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.855042335754129e-06,
+      "loss": 0.1285,
+      "step": 3381
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.846690410686025e-06,
+      "loss": 0.0895,
+      "step": 3382
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.838342071755049e-06,
+      "loss": 0.0571,
+      "step": 3383
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.829997322633129e-06,
+      "loss": 0.1372,
+      "step": 3384
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.821656166990606e-06,
+      "loss": 0.1205,
+      "step": 3385
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.813318608496237e-06,
+      "loss": 0.1093,
+      "step": 3386
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.8049846508171954e-06,
+      "loss": 0.1186,
+      "step": 3387
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.79665429761908e-06,
+      "loss": 0.0952,
+      "step": 3388
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.788327552565888e-06,
+      "loss": 0.123,
+      "step": 3389
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.780004419320059e-06,
+      "loss": 0.1553,
+      "step": 3390
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.771684901542403e-06,
+      "loss": 0.1022,
+      "step": 3391
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.763369002892177e-06,
+      "loss": 0.1178,
+      "step": 3392
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.7550567270270265e-06,
+      "loss": 0.1173,
+      "step": 3393
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.74674807760301e-06,
+      "loss": 0.1266,
+      "step": 3394
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.738443058274579e-06,
+      "loss": 0.078,
+      "step": 3395
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.730141672694619e-06,
+      "loss": 0.1344,
+      "step": 3396
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.721843924514374e-06,
+      "loss": 0.1136,
+      "step": 3397
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.713549817383524e-06,
+      "loss": 0.1419,
+      "step": 3398
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.705259354950133e-06,
+      "loss": 0.1176,
+      "step": 3399
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.6969725408606595e-06,
+      "loss": 0.1048,
+      "step": 3400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.688689378759965e-06,
+      "loss": 0.1545,
+      "step": 3401
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.680409872291298e-06,
+      "loss": 0.1294,
+      "step": 3402
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.672134025096296e-06,
+      "loss": 0.1066,
+      "step": 3403
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.6638618408150095e-06,
+      "loss": 0.0725,
+      "step": 3404
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.6555933230858424e-06,
+      "loss": 0.1287,
+      "step": 3405
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.647328475545611e-06,
+      "loss": 0.0725,
+      "step": 3406
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.639067301829526e-06,
+      "loss": 0.1281,
+      "step": 3407
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.630809805571145e-06,
+      "loss": 0.1365,
+      "step": 3408
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.622555990402443e-06,
+      "loss": 0.0932,
+      "step": 3409
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.614305859953762e-06,
+      "loss": 0.1222,
+      "step": 3410
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.6060594178538226e-06,
+      "loss": 0.1127,
+      "step": 3411
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.597816667729727e-06,
+      "loss": 0.0856,
+      "step": 3412
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.58957761320695e-06,
+      "loss": 0.0894,
+      "step": 3413
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.581342257909337e-06,
+      "loss": 0.0855,
+      "step": 3414
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.573110605459129e-06,
+      "loss": 0.0939,
+      "step": 3415
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.5648826594769e-06,
+      "loss": 0.1121,
+      "step": 3416
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.55665842358163e-06,
+      "loss": 0.0848,
+      "step": 3417
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.548437901390649e-06,
+      "loss": 0.096,
+      "step": 3418
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.5402210965196555e-06,
+      "loss": 0.0831,
+      "step": 3419
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.532008012582707e-06,
+      "loss": 0.1381,
+      "step": 3420
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.523798653192253e-06,
+      "loss": 0.1056,
+      "step": 3421
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.5155930219590555e-06,
+      "loss": 0.0876,
+      "step": 3422
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.507391122492287e-06,
+      "loss": 0.1074,
+      "step": 3423
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.499192958399446e-06,
+      "loss": 0.1082,
+      "step": 3424
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.490998533286401e-06,
+      "loss": 0.1283,
+      "step": 3425
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.4828078507573714e-06,
+      "loss": 0.0775,
+      "step": 3426
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.474620914414935e-06,
+      "loss": 0.1094,
+      "step": 3427
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.466437727860013e-06,
+      "loss": 0.1311,
+      "step": 3428
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.458258294691898e-06,
+      "loss": 0.098,
+      "step": 3429
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.4500826185081975e-06,
+      "loss": 0.1004,
+      "step": 3430
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.4419107029049016e-06,
+      "loss": 0.1116,
+      "step": 3431
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.433742551476326e-06,
+      "loss": 0.1058,
+      "step": 3432
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.425578167815136e-06,
+      "loss": 0.0937,
+      "step": 3433
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.417417555512335e-06,
+      "loss": 0.1109,
+      "step": 3434
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.409260718157287e-06,
+      "loss": 0.1223,
+      "step": 3435
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.401107659337659e-06,
+      "loss": 0.1245,
+      "step": 3436
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.392958382639494e-06,
+      "loss": 0.1091,
+      "step": 3437
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.384812891647153e-06,
+      "loss": 0.0905,
+      "step": 3438
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.3766711899433235e-06,
+      "loss": 0.0816,
+      "step": 3439
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.368533281109059e-06,
+      "loss": 0.1342,
+      "step": 3440
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.360399168723695e-06,
+      "loss": 0.0874,
+      "step": 3441
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.3522688563649465e-06,
+      "loss": 0.1035,
+      "step": 3442
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.344142347608828e-06,
+      "loss": 0.1064,
+      "step": 3443
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.336019646029688e-06,
+      "loss": 0.1288,
+      "step": 3444
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.327900755200197e-06,
+      "loss": 0.0819,
+      "step": 3445
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.31978567869137e-06,
+      "loss": 0.0614,
+      "step": 3446
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.311674420072503e-06,
+      "loss": 0.1045,
+      "step": 3447
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.303566982911258e-06,
+      "loss": 0.1172,
+      "step": 3448
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.295463370773586e-06,
+      "loss": 0.1154,
+      "step": 3449
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.287363587223769e-06,
+      "loss": 0.1188,
+      "step": 3450
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.279267635824401e-06,
+      "loss": 0.0816,
+      "step": 3451
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.271175520136387e-06,
+      "loss": 0.0748,
+      "step": 3452
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.263087243718947e-06,
+      "loss": 0.0899,
+      "step": 3453
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.25500281012963e-06,
+      "loss": 0.1054,
+      "step": 3454
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.246922222924257e-06,
+      "loss": 0.1423,
+      "step": 3455
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.238845485656992e-06,
+      "loss": 0.085,
+      "step": 3456
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.230772601880292e-06,
+      "loss": 0.1293,
+      "step": 3457
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.222703575144918e-06,
+      "loss": 0.0922,
+      "step": 3458
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.214638408999929e-06,
+      "loss": 0.1176,
+      "step": 3459
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.206577106992714e-06,
+      "loss": 0.1107,
+      "step": 3460
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.198519672668916e-06,
+      "loss": 0.1369,
+      "step": 3461
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.19046610957252e-06,
+      "loss": 0.0772,
+      "step": 3462
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.1824164212457866e-06,
+      "loss": 0.129,
+      "step": 3463
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.174370611229274e-06,
+      "loss": 0.0896,
+      "step": 3464
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.1663286830618365e-06,
+      "loss": 0.101,
+      "step": 3465
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.15829064028062e-06,
+      "loss": 0.1002,
+      "step": 3466
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.150256486421071e-06,
+      "loss": 0.1004,
+      "step": 3467
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.142226225016911e-06,
+      "loss": 0.1199,
+      "step": 3468
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.1341998596001585e-06,
+      "loss": 0.0844,
+      "step": 3469
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.126177393701108e-06,
+      "loss": 0.1527,
+      "step": 3470
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.118158830848366e-06,
+      "loss": 0.1042,
+      "step": 3471
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.110144174568777e-06,
+      "loss": 0.1093,
+      "step": 3472
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.102133428387511e-06,
+      "loss": 0.1125,
+      "step": 3473
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.094126595827997e-06,
+      "loss": 0.0797,
+      "step": 3474
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.086123680411943e-06,
+      "loss": 0.1327,
+      "step": 3475
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.078124685659339e-06,
+      "loss": 0.1001,
+      "step": 3476
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.070129615088448e-06,
+      "loss": 0.1226,
+      "step": 3477
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.062138472215801e-06,
+      "loss": 0.0677,
+      "step": 3478
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.054151260556226e-06,
+      "loss": 0.119,
+      "step": 3479
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.04616798362278e-06,
+      "loss": 0.087,
+      "step": 3480
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.038188644926834e-06,
+      "loss": 0.116,
+      "step": 3481
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.030213247977995e-06,
+      "loss": 0.1063,
+      "step": 3482
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.022241796284152e-06,
+      "loss": 0.127,
+      "step": 3483
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.014274293351446e-06,
+      "loss": 0.0967,
+      "step": 3484
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.006310742684306e-06,
+      "loss": 0.0604,
+      "step": 3485
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.998351147785383e-06,
+      "loss": 0.09,
+      "step": 3486
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.990395512155631e-06,
+      "loss": 0.0807,
+      "step": 3487
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.982443839294235e-06,
+      "loss": 0.0654,
+      "step": 3488
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.974496132698642e-06,
+      "loss": 0.1019,
+      "step": 3489
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.966552395864563e-06,
+      "loss": 0.1193,
+      "step": 3490
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.958612632285952e-06,
+      "loss": 0.1401,
+      "step": 3491
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.950676845455016e-06,
+      "loss": 0.0912,
+      "step": 3492
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.942745038862235e-06,
+      "loss": 0.0688,
+      "step": 3493
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.934817215996297e-06,
+      "loss": 0.0917,
+      "step": 3494
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.926893380344179e-06,
+      "loss": 0.0956,
+      "step": 3495
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.918973535391083e-06,
+      "loss": 0.1035,
+      "step": 3496
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.91105768462045e-06,
+      "loss": 0.1027,
+      "step": 3497
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.903145831513989e-06,
+      "loss": 0.1141,
+      "step": 3498
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.895237979551627e-06,
+      "loss": 0.1005,
+      "step": 3499
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.887334132211541e-06,
+      "loss": 0.0812,
+      "step": 3500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.879434292970142e-06,
+      "loss": 0.1213,
+      "step": 3501
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.871538465302085e-06,
+      "loss": 0.095,
+      "step": 3502
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.86364665268025e-06,
+      "loss": 0.1083,
+      "step": 3503
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.855758858575773e-06,
+      "loss": 0.1257,
+      "step": 3504
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.847875086457987e-06,
+      "loss": 0.1044,
+      "step": 3505
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.839995339794492e-06,
+      "loss": 0.0846,
+      "step": 3506
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.832119622051095e-06,
+      "loss": 0.0731,
+      "step": 3507
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.824247936691841e-06,
+      "loss": 0.1274,
+      "step": 3508
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.816380287178989e-06,
+      "loss": 0.1038,
+      "step": 3509
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.808516676973051e-06,
+      "loss": 0.1049,
+      "step": 3510
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.80065710953272e-06,
+      "loss": 0.1064,
+      "step": 3511
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.792801588314948e-06,
+      "loss": 0.13,
+      "step": 3512
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.784950116774893e-06,
+      "loss": 0.1073,
+      "step": 3513
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.77710269836593e-06,
+      "loss": 0.1236,
+      "step": 3514
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.769259336539654e-06,
+      "loss": 0.1152,
+      "step": 3515
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.7614200347458736e-06,
+      "loss": 0.1476,
+      "step": 3516
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.753584796432608e-06,
+      "loss": 0.0892,
+      "step": 3517
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.745753625046113e-06,
+      "loss": 0.119,
+      "step": 3518
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.737926524030814e-06,
+      "loss": 0.0823,
+      "step": 3519
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.730103496829385e-06,
+      "loss": 0.1125,
+      "step": 3520
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.722284546882687e-06,
+      "loss": 0.1241,
+      "step": 3521
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.714469677629794e-06,
+      "loss": 0.1546,
+      "step": 3522
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.706658892507979e-06,
+      "loss": 0.1105,
+      "step": 3523
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.698852194952737e-06,
+      "loss": 0.0775,
+      "step": 3524
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.691049588397733e-06,
+      "loss": 0.077,
+      "step": 3525
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.6832510762748675e-06,
+      "loss": 0.092,
+      "step": 3526
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.675456662014218e-06,
+      "loss": 0.1105,
+      "step": 3527
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.66766634904406e-06,
+      "loss": 0.1411,
+      "step": 3528
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.659880140790888e-06,
+      "loss": 0.0669,
+      "step": 3529
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.652098040679354e-06,
+      "loss": 0.097,
+      "step": 3530
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.644320052132336e-06,
+      "loss": 0.1131,
+      "step": 3531
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.636546178570886e-06,
+      "loss": 0.0698,
+      "step": 3532
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.6287764234142516e-06,
+      "loss": 0.1263,
+      "step": 3533
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.621010790079862e-06,
+      "loss": 0.0849,
+      "step": 3534
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.613249281983357e-06,
+      "loss": 0.1282,
+      "step": 3535
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.605491902538521e-06,
+      "loss": 0.1061,
+      "step": 3536
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.5977386551573626e-06,
+      "loss": 0.129,
+      "step": 3537
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.5899895432500515e-06,
+      "loss": 0.1334,
+      "step": 3538
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.582244570224942e-06,
+      "loss": 0.0903,
+      "step": 3539
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.574503739488571e-06,
+      "loss": 0.1304,
+      "step": 3540
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.56676705444565e-06,
+      "loss": 0.1185,
+      "step": 3541
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.5590345184990664e-06,
+      "loss": 0.0996,
+      "step": 3542
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.551306135049897e-06,
+      "loss": 0.1,
+      "step": 3543
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.543581907497362e-06,
+      "loss": 0.0978,
+      "step": 3544
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.535861839238885e-06,
+      "loss": 0.0925,
+      "step": 3545
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.528145933670044e-06,
+      "loss": 0.0982,
+      "step": 3546
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.5204341941845895e-06,
+      "loss": 0.1074,
+      "step": 3547
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.512726624174433e-06,
+      "loss": 0.0812,
+      "step": 3548
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.505023227029671e-06,
+      "loss": 0.164,
+      "step": 3549
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.497324006138536e-06,
+      "loss": 0.0797,
+      "step": 3550
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.489628964887453e-06,
+      "loss": 0.1035,
+      "step": 3551
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.4819381066609906e-06,
+      "loss": 0.1149,
+      "step": 3552
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.4742514348418805e-06,
+      "loss": 0.1048,
+      "step": 3553
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.466568952811017e-06,
+      "loss": 0.1201,
+      "step": 3554
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.458890663947449e-06,
+      "loss": 0.1182,
+      "step": 3555
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.451216571628374e-06,
+      "loss": 0.1219,
+      "step": 3556
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.443546679229162e-06,
+      "loss": 0.0825,
+      "step": 3557
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.435880990123323e-06,
+      "loss": 0.1245,
+      "step": 3558
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.428219507682511e-06,
+      "loss": 0.0885,
+      "step": 3559
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.420562235276555e-06,
+      "loss": 0.1058,
+      "step": 3560
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.412909176273397e-06,
+      "loss": 0.1074,
+      "step": 3561
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.405260334039158e-06,
+      "loss": 0.0931,
+      "step": 3562
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.397615711938088e-06,
+      "loss": 0.1063,
+      "step": 3563
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.389975313332584e-06,
+      "loss": 0.1033,
+      "step": 3564
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.3823391415831836e-06,
+      "loss": 0.1493,
+      "step": 3565
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.374707200048568e-06,
+      "loss": 0.1002,
+      "step": 3566
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.367079492085551e-06,
+      "loss": 0.1222,
+      "step": 3567
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.359456021049105e-06,
+      "loss": 0.1036,
+      "step": 3568
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.351836790292305e-06,
+      "loss": 0.088,
+      "step": 3569
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.3442218031663945e-06,
+      "loss": 0.0786,
+      "step": 3570
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.336611063020732e-06,
+      "loss": 0.0907,
+      "step": 3571
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.329004573202809e-06,
+      "loss": 0.0928,
+      "step": 3572
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.321402337058246e-06,
+      "loss": 0.106,
+      "step": 3573
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.313804357930814e-06,
+      "loss": 0.0985,
+      "step": 3574
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.306210639162374e-06,
+      "loss": 0.0816,
+      "step": 3575
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.2986211840929485e-06,
+      "loss": 0.096,
+      "step": 3576
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.291035996060662e-06,
+      "loss": 0.0995,
+      "step": 3577
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.283455078401775e-06,
+      "loss": 0.1209,
+      "step": 3578
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.275878434450661e-06,
+      "loss": 0.1488,
+      "step": 3579
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.26830606753982e-06,
+      "loss": 0.0952,
+      "step": 3580
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.2607379809998635e-06,
+      "loss": 0.0957,
+      "step": 3581
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.253174178159536e-06,
+      "loss": 0.0867,
+      "step": 3582
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.245614662345672e-06,
+      "loss": 0.0848,
+      "step": 3583
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.238059436883248e-06,
+      "loss": 0.0763,
+      "step": 3584
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.230508505095337e-06,
+      "loss": 0.112,
+      "step": 3585
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.222961870303127e-06,
+      "loss": 0.1514,
+      "step": 3586
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.21541953582591e-06,
+      "loss": 0.1085,
+      "step": 3587
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.207881504981104e-06,
+      "loss": 0.0765,
+      "step": 3588
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.200347781084218e-06,
+      "loss": 0.1187,
+      "step": 3589
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.1928183674488714e-06,
+      "loss": 0.1102,
+      "step": 3590
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.185293267386789e-06,
+      "loss": 0.0881,
+      "step": 3591
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.177772484207792e-06,
+      "loss": 0.0931,
+      "step": 3592
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.170256021219823e-06,
+      "loss": 0.1039,
+      "step": 3593
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.162743881728888e-06,
+      "loss": 0.1092,
+      "step": 3594
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.155236069039132e-06,
+      "loss": 0.1363,
+      "step": 3595
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.147732586452769e-06,
+      "loss": 0.0856,
+      "step": 3596
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.140233437270119e-06,
+      "loss": 0.0924,
+      "step": 3597
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.132738624789591e-06,
+      "loss": 0.093,
+      "step": 3598
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.125248152307702e-06,
+      "loss": 0.1104,
+      "step": 3599
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.11776202311903e-06,
+      "loss": 0.114,
+      "step": 3600
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.1102802405162745e-06,
+      "loss": 0.0979,
+      "step": 3601
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.102802807790205e-06,
+      "loss": 0.0964,
+      "step": 3602
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.0953297282296836e-06,
+      "loss": 0.0758,
+      "step": 3603
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.0878610051216525e-06,
+      "loss": 0.1578,
+      "step": 3604
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.080396641751145e-06,
+      "loss": 0.0789,
+      "step": 3605
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.072936641401269e-06,
+      "loss": 0.1224,
+      "step": 3606
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.065481007353228e-06,
+      "loss": 0.0971,
+      "step": 3607
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.05802974288628e-06,
+      "loss": 0.1033,
+      "step": 3608
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.050582851277787e-06,
+      "loss": 0.1001,
+      "step": 3609
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.043140335803174e-06,
+      "loss": 0.0756,
+      "step": 3610
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.035702199735941e-06,
+      "loss": 0.0698,
+      "step": 3611
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.028268446347664e-06,
+      "loss": 0.1488,
+      "step": 3612
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.020839078908003e-06,
+      "loss": 0.1242,
+      "step": 3613
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.0134141006846585e-06,
+      "loss": 0.0827,
+      "step": 3614
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.005993514943435e-06,
+      "loss": 0.125,
+      "step": 3615
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.998577324948185e-06,
+      "loss": 0.1191,
+      "step": 3616
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.991165533960828e-06,
+      "loss": 0.0872,
+      "step": 3617
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.983758145241365e-06,
+      "loss": 0.0913,
+      "step": 3618
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.976355162047833e-06,
+      "loss": 0.0908,
+      "step": 3619
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.968956587636357e-06,
+      "loss": 0.1142,
+      "step": 3620
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.961562425261112e-06,
+      "loss": 0.1012,
+      "step": 3621
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.9541726781743305e-06,
+      "loss": 0.0883,
+      "step": 3622
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.946787349626302e-06,
+      "loss": 0.1415,
+      "step": 3623
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.93940644286539e-06,
+      "loss": 0.1056,
+      "step": 3624
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.93202996113798e-06,
+      "loss": 0.1044,
+      "step": 3625
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.9246579076885475e-06,
+      "loss": 0.0871,
+      "step": 3626
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.917290285759594e-06,
+      "loss": 0.0951,
+      "step": 3627
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.909927098591686e-06,
+      "loss": 0.0888,
+      "step": 3628
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.902568349423431e-06,
+      "loss": 0.0954,
+      "step": 3629
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.89521404149149e-06,
+      "loss": 0.1327,
+      "step": 3630
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.887864178030563e-06,
+      "loss": 0.0901,
+      "step": 3631
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.880518762273416e-06,
+      "loss": 0.1071,
+      "step": 3632
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.8731777974508214e-06,
+      "loss": 0.107,
+      "step": 3633
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.865841286791636e-06,
+      "loss": 0.1157,
+      "step": 3634
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.8585092335227295e-06,
+      "loss": 0.0768,
+      "step": 3635
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.851181640869018e-06,
+      "loss": 0.1047,
+      "step": 3636
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.843858512053453e-06,
+      "loss": 0.1078,
+      "step": 3637
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.836539850297042e-06,
+      "loss": 0.0704,
+      "step": 3638
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.829225658818792e-06,
+      "loss": 0.1039,
+      "step": 3639
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.821915940835778e-06,
+      "loss": 0.0909,
+      "step": 3640
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.814610699563089e-06,
+      "loss": 0.0984,
+      "step": 3641
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.8073099382138495e-06,
+      "loss": 0.1423,
+      "step": 3642
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.800013659999211e-06,
+      "loss": 0.0947,
+      "step": 3643
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.792721868128359e-06,
+      "loss": 0.1528,
+      "step": 3644
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.785434565808496e-06,
+      "loss": 0.1246,
+      "step": 3645
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.778151756244871e-06,
+      "loss": 0.0989,
+      "step": 3646
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.770873442640721e-06,
+      "loss": 0.1042,
+      "step": 3647
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.763599628197337e-06,
+      "loss": 0.0667,
+      "step": 3648
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.75633031611403e-06,
+      "loss": 0.095,
+      "step": 3649
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.749065509588103e-06,
+      "loss": 0.1232,
+      "step": 3650
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.741805211814909e-06,
+      "loss": 0.102,
+      "step": 3651
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.734549425987801e-06,
+      "loss": 0.0887,
+      "step": 3652
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.727298155298148e-06,
+      "loss": 0.0844,
+      "step": 3653
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.720051402935338e-06,
+      "loss": 0.1025,
+      "step": 3654
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.712809172086771e-06,
+      "loss": 0.0877,
+      "step": 3655
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.705571465937849e-06,
+      "loss": 0.1117,
+      "step": 3656
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.698338287672009e-06,
+      "loss": 0.0861,
+      "step": 3657
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.691109640470657e-06,
+      "loss": 0.1306,
+      "step": 3658
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.683885527513244e-06,
+      "loss": 0.0755,
+      "step": 3659
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.676665951977208e-06,
+      "loss": 0.1218,
+      "step": 3660
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.6694509170379925e-06,
+      "loss": 0.0864,
+      "step": 3661
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.66224042586904e-06,
+      "loss": 0.1256,
+      "step": 3662
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.655034481641816e-06,
+      "loss": 0.0774,
+      "step": 3663
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.647833087525751e-06,
+      "loss": 0.1141,
+      "step": 3664
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.640636246688305e-06,
+      "loss": 0.0912,
+      "step": 3665
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.633443962294923e-06,
+      "loss": 0.1305,
+      "step": 3666
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.626256237509044e-06,
+      "loss": 0.1073,
+      "step": 3667
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.6190730754921035e-06,
+      "loss": 0.0972,
+      "step": 3668
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.611894479403533e-06,
+      "loss": 0.0984,
+      "step": 3669
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.604720452400747e-06,
+      "loss": 0.0726,
+      "step": 3670
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.59755099763917e-06,
+      "loss": 0.1191,
+      "step": 3671
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.590386118272186e-06,
+      "loss": 0.0998,
+      "step": 3672
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.5832258174511945e-06,
+      "loss": 0.1087,
+      "step": 3673
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.576070098325566e-06,
+      "loss": 0.0586,
+      "step": 3674
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.5689189640426585e-06,
+      "loss": 0.094,
+      "step": 3675
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.56177241774781e-06,
+      "loss": 0.0908,
+      "step": 3676
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.5546304625843606e-06,
+      "loss": 0.1162,
+      "step": 3677
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.5474931016935924e-06,
+      "loss": 0.0614,
+      "step": 3678
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.540360338214805e-06,
+      "loss": 0.0595,
+      "step": 3679
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.5332321752852566e-06,
+      "loss": 0.145,
+      "step": 3680
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.526108616040181e-06,
+      "loss": 0.0879,
+      "step": 3681
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.518989663612804e-06,
+      "loss": 0.0958,
+      "step": 3682
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.511875321134298e-06,
+      "loss": 0.1178,
+      "step": 3683
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.504765591733833e-06,
+      "loss": 0.0847,
+      "step": 3684
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.497660478538535e-06,
+      "loss": 0.0924,
+      "step": 3685
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.490559984673507e-06,
+      "loss": 0.1188,
+      "step": 3686
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4834641132618095e-06,
+      "loss": 0.0996,
+      "step": 3687
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4763728674244965e-06,
+      "loss": 0.0844,
+      "step": 3688
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.469286250280546e-06,
+      "loss": 0.0934,
+      "step": 3689
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.46220426494694e-06,
+      "loss": 0.11,
+      "step": 3690
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.455126914538599e-06,
+      "loss": 0.0733,
+      "step": 3691
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.448054202168413e-06,
+      "loss": 0.0811,
+      "step": 3692
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4409861309472325e-06,
+      "loss": 0.0902,
+      "step": 3693
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.433922703983865e-06,
+      "loss": 0.0655,
+      "step": 3694
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4268639243850696e-06,
+      "loss": 0.1066,
+      "step": 3695
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.419809795255583e-06,
+      "loss": 0.0898,
+      "step": 3696
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.412760319698061e-06,
+      "loss": 0.1094,
+      "step": 3697
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.405715500813146e-06,
+      "loss": 0.1175,
+      "step": 3698
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.398675341699416e-06,
+      "loss": 0.0824,
+      "step": 3699
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.391639845453398e-06,
+      "loss": 0.0741,
+      "step": 3700
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.3846090151695714e-06,
+      "loss": 0.1019,
+      "step": 3701
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.377582853940376e-06,
+      "loss": 0.105,
+      "step": 3702
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.370561364856168e-06,
+      "loss": 0.1125,
+      "step": 3703
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.3635445510052806e-06,
+      "loss": 0.1005,
+      "step": 3704
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.356532415473972e-06,
+      "loss": 0.1188,
+      "step": 3705
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.349524961346447e-06,
+      "loss": 0.1192,
+      "step": 3706
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.3425221917048534e-06,
+      "loss": 0.1053,
+      "step": 3707
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.335524109629269e-06,
+      "loss": 0.0989,
+      "step": 3708
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.3285307181977305e-06,
+      "loss": 0.0908,
+      "step": 3709
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.321542020486191e-06,
+      "loss": 0.0932,
+      "step": 3710
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.314558019568549e-06,
+      "loss": 0.1124,
+      "step": 3711
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.30757871851663e-06,
+      "loss": 0.07,
+      "step": 3712
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.300604120400211e-06,
+      "loss": 0.0814,
+      "step": 3713
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.293634228286967e-06,
+      "loss": 0.0887,
+      "step": 3714
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.28666904524254e-06,
+      "loss": 0.0622,
+      "step": 3715
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.279708574330476e-06,
+      "loss": 0.0853,
+      "step": 3716
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.272752818612256e-06,
+      "loss": 0.0982,
+      "step": 3717
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.26580178114729e-06,
+      "loss": 0.0899,
+      "step": 3718
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.258855464992907e-06,
+      "loss": 0.0818,
+      "step": 3719
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2519138732043574e-06,
+      "loss": 0.089,
+      "step": 3720
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.244977008834837e-06,
+      "loss": 0.095,
+      "step": 3721
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.238044874935421e-06,
+      "loss": 0.1204,
+      "step": 3722
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.231117474555141e-06,
+      "loss": 0.0794,
+      "step": 3723
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.22419481074093e-06,
+      "loss": 0.1098,
+      "step": 3724
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2172768865376386e-06,
+      "loss": 0.091,
+      "step": 3725
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2103637049880275e-06,
+      "loss": 0.0719,
+      "step": 3726
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.203455269132797e-06,
+      "loss": 0.1161,
+      "step": 3727
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.196551582010519e-06,
+      "loss": 0.126,
+      "step": 3728
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.189652646657714e-06,
+      "loss": 0.0875,
+      "step": 3729
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.182758466108791e-06,
+      "loss": 0.0872,
+      "step": 3730
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.175869043396079e-06,
+      "loss": 0.1139,
+      "step": 3731
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.1689843815498025e-06,
+      "loss": 0.1198,
+      "step": 3732
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.162104483598105e-06,
+      "loss": 0.1305,
+      "step": 3733
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.15522935256702e-06,
+      "loss": 0.106,
+      "step": 3734
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.148358991480509e-06,
+      "loss": 0.1094,
+      "step": 3735
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.141493403360399e-06,
+      "loss": 0.0869,
+      "step": 3736
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.134632591226453e-06,
+      "loss": 0.0663,
+      "step": 3737
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.127776558096312e-06,
+      "loss": 0.1198,
+      "step": 3738
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.120925306985521e-06,
+      "loss": 0.084,
+      "step": 3739
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.114078840907528e-06,
+      "loss": 0.1128,
+      "step": 3740
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.1072371628736655e-06,
+      "loss": 0.0983,
+      "step": 3741
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.100400275893166e-06,
+      "loss": 0.0573,
+      "step": 3742
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.093568182973153e-06,
+      "loss": 0.095,
+      "step": 3743
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.086740887118644e-06,
+      "loss": 0.1226,
+      "step": 3744
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.079918391332537e-06,
+      "loss": 0.1054,
+      "step": 3745
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.073100698615642e-06,
+      "loss": 0.0823,
+      "step": 3746
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.066287811966623e-06,
+      "loss": 0.0819,
+      "step": 3747
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.05947973438206e-06,
+      "loss": 0.107,
+      "step": 3748
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0526764688564045e-06,
+      "loss": 0.1154,
+      "step": 3749
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0458780183819875e-06,
+      "loss": 0.1572,
+      "step": 3750
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0390843859490255e-06,
+      "loss": 0.0837,
+      "step": 3751
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.032295574545631e-06,
+      "loss": 0.1136,
+      "step": 3752
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.025511587157766e-06,
+      "loss": 0.0815,
+      "step": 3753
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0187324267693e-06,
+      "loss": 0.0789,
+      "step": 3754
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.01195809636196e-06,
+      "loss": 0.1044,
+      "step": 3755
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.005188598915359e-06,
+      "loss": 0.1337,
+      "step": 3756
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.998423937406979e-06,
+      "loss": 0.1214,
+      "step": 3757
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.991664114812175e-06,
+      "loss": 0.0989,
+      "step": 3758
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.984909134104175e-06,
+      "loss": 0.1177,
+      "step": 3759
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.978158998254086e-06,
+      "loss": 0.0985,
+      "step": 3760
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.971413710230862e-06,
+      "loss": 0.1049,
+      "step": 3761
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.9646732730013504e-06,
+      "loss": 0.1133,
+      "step": 3762
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.957937689530248e-06,
+      "loss": 0.0955,
+      "step": 3763
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.951206962780123e-06,
+      "loss": 0.0926,
+      "step": 3764
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.944481095711401e-06,
+      "loss": 0.0651,
+      "step": 3765
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.937760091282387e-06,
+      "loss": 0.0879,
+      "step": 3766
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.931043952449222e-06,
+      "loss": 0.0978,
+      "step": 3767
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.92433268216593e-06,
+      "loss": 0.0881,
+      "step": 3768
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.917626283384381e-06,
+      "loss": 0.1318,
+      "step": 3769
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.910924759054301e-06,
+      "loss": 0.1404,
+      "step": 3770
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.904228112123292e-06,
+      "loss": 0.1099,
+      "step": 3771
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.897536345536774e-06,
+      "loss": 0.1509,
+      "step": 3772
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.890849462238057e-06,
+      "loss": 0.1018,
+      "step": 3773
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.884167465168283e-06,
+      "loss": 0.0865,
+      "step": 3774
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.877490357266448e-06,
+      "loss": 0.0935,
+      "step": 3775
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.870818141469402e-06,
+      "loss": 0.0922,
+      "step": 3776
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.864150820711839e-06,
+      "loss": 0.09,
+      "step": 3777
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.857488397926298e-06,
+      "loss": 0.1075,
+      "step": 3778
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8508308760431735e-06,
+      "loss": 0.0805,
+      "step": 3779
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.844178257990696e-06,
+      "loss": 0.1157,
+      "step": 3780
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.83753054669494e-06,
+      "loss": 0.0829,
+      "step": 3781
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.830887745079825e-06,
+      "loss": 0.0767,
+      "step": 3782
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.824249856067108e-06,
+      "loss": 0.1252,
+      "step": 3783
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.81761688257638e-06,
+      "loss": 0.0701,
+      "step": 3784
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.81098882752509e-06,
+      "loss": 0.1055,
+      "step": 3785
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8043656938284947e-06,
+      "loss": 0.0997,
+      "step": 3786
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.797747484399712e-06,
+      "loss": 0.058,
+      "step": 3787
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7911342021496802e-06,
+      "loss": 0.1463,
+      "step": 3788
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7845258499871737e-06,
+      "loss": 0.0967,
+      "step": 3789
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.777922430818792e-06,
+      "loss": 0.0848,
+      "step": 3790
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.771323947548987e-06,
+      "loss": 0.1268,
+      "step": 3791
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7647304030800052e-06,
+      "loss": 0.0949,
+      "step": 3792
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7581418003119517e-06,
+      "loss": 0.0925,
+      "step": 3793
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7515581421427435e-06,
+      "loss": 0.136,
+      "step": 3794
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.744979431468123e-06,
+      "loss": 0.1074,
+      "step": 3795
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7384056711816585e-06,
+      "loss": 0.1069,
+      "step": 3796
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.731836864174742e-06,
+      "loss": 0.123,
+      "step": 3797
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.7252730133365825e-06,
+      "loss": 0.1077,
+      "step": 3798
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.718714121554219e-06,
+      "loss": 0.0666,
+      "step": 3799
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.712160191712499e-06,
+      "loss": 0.0582,
+      "step": 3800
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.705611226694092e-06,
+      "loss": 0.1069,
+      "step": 3801
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6990672293794825e-06,
+      "loss": 0.1059,
+      "step": 3802
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6925282026469674e-06,
+      "loss": 0.0761,
+      "step": 3803
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.685994149372668e-06,
+      "loss": 0.0673,
+      "step": 3804
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6794650724305062e-06,
+      "loss": 0.0865,
+      "step": 3805
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6729409746922205e-06,
+      "loss": 0.0831,
+      "step": 3806
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.666421859027357e-06,
+      "loss": 0.0982,
+      "step": 3807
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6599077283032733e-06,
+      "loss": 0.1195,
+      "step": 3808
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.653398585385128e-06,
+      "loss": 0.0878,
+      "step": 3809
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.646894433135903e-06,
+      "loss": 0.1072,
+      "step": 3810
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.640395274416359e-06,
+      "loss": 0.0857,
+      "step": 3811
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.633901112085083e-06,
+      "loss": 0.1296,
+      "step": 3812
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6274119489984536e-06,
+      "loss": 0.0446,
+      "step": 3813
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6209277880106543e-06,
+      "loss": 0.1164,
+      "step": 3814
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6144486319736644e-06,
+      "loss": 0.1279,
+      "step": 3815
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6079744837372646e-06,
+      "loss": 0.0714,
+      "step": 3816
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.60150534614903e-06,
+      "loss": 0.0888,
+      "step": 3817
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.5950412220543407e-06,
+      "loss": 0.0774,
+      "step": 3818
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.5885821142963637e-06,
+      "loss": 0.0932,
+      "step": 3819
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.5821280257160604e-06,
+      "loss": 0.1182,
+      "step": 3820
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.575678959152185e-06,
+      "loss": 0.0908,
+      "step": 3821
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.569234917441283e-06,
+      "loss": 0.0842,
+      "step": 3822
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5627959034176877e-06,
+      "loss": 0.1093,
+      "step": 3823
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5563619199135365e-06,
+      "loss": 0.0789,
+      "step": 3824
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.549932969758723e-06,
+      "loss": 0.133,
+      "step": 3825
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5435090557809563e-06,
+      "loss": 0.1065,
+      "step": 3826
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5370901808057184e-06,
+      "loss": 0.0899,
+      "step": 3827
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5306763476562703e-06,
+      "loss": 0.0906,
+      "step": 3828
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.524267559153668e-06,
+      "loss": 0.0671,
+      "step": 3829
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.517863818116739e-06,
+      "loss": 0.049,
+      "step": 3830
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.511465127362092e-06,
+      "loss": 0.0711,
+      "step": 3831
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.505071489704117e-06,
+      "loss": 0.1171,
+      "step": 3832
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4986829079549797e-06,
+      "loss": 0.1356,
+      "step": 3833
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.492299384924618e-06,
+      "loss": 0.1099,
+      "step": 3834
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.485920923420763e-06,
+      "loss": 0.1001,
+      "step": 3835
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.479547526248888e-06,
+      "loss": 0.0852,
+      "step": 3836
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4731791962122694e-06,
+      "loss": 0.1013,
+      "step": 3837
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.466815936111937e-06,
+      "loss": 0.0635,
+      "step": 3838
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4604577487466974e-06,
+      "loss": 0.1065,
+      "step": 3839
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4541046369131235e-06,
+      "loss": 0.1192,
+      "step": 3840
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4477566034055573e-06,
+      "loss": 0.0982,
+      "step": 3841
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.441413651016101e-06,
+      "loss": 0.1154,
+      "step": 3842
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4350757825346374e-06,
+      "loss": 0.1184,
+      "step": 3843
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.428743000748797e-06,
+      "loss": 0.1136,
+      "step": 3844
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.422415308443983e-06,
+      "loss": 0.0955,
+      "step": 3845
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.416092708403351e-06,
+      "loss": 0.1299,
+      "step": 3846
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4097752034078285e-06,
+      "loss": 0.1194,
+      "step": 3847
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4034627962360853e-06,
+      "loss": 0.0686,
+      "step": 3848
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3971554896645745e-06,
+      "loss": 0.0989,
+      "step": 3849
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.390853286467475e-06,
+      "loss": 0.0766,
+      "step": 3850
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3845561894167482e-06,
+      "loss": 0.1113,
+      "step": 3851
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.378264201282093e-06,
+      "loss": 0.0807,
+      "step": 3852
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.371977324830966e-06,
+      "loss": 0.109,
+      "step": 3853
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3656955628285787e-06,
+      "loss": 0.0903,
+      "step": 3854
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3594189180378854e-06,
+      "loss": 0.1023,
+      "step": 3855
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.353147393219595e-06,
+      "loss": 0.0717,
+      "step": 3856
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3468809911321675e-06,
+      "loss": 0.1025,
+      "step": 3857
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3406197145318062e-06,
+      "loss": 0.107,
+      "step": 3858
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3343635661724538e-06,
+      "loss": 0.0869,
+      "step": 3859
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.328112548805815e-06,
+      "loss": 0.0828,
+      "step": 3860
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3218666651813095e-06,
+      "loss": 0.0712,
+      "step": 3861
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3156259180461297e-06,
+      "loss": 0.0998,
+      "step": 3862
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3093903101451893e-06,
+      "loss": 0.1246,
+      "step": 3863
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3031598442211468e-06,
+      "loss": 0.1015,
+      "step": 3864
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2969345230144014e-06,
+      "loss": 0.1109,
+      "step": 3865
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2907143492630846e-06,
+      "loss": 0.0936,
+      "step": 3866
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2844993257030632e-06,
+      "loss": 0.0776,
+      "step": 3867
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2782894550679515e-06,
+      "loss": 0.071,
+      "step": 3868
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2720847400890817e-06,
+      "loss": 0.087,
+      "step": 3869
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2658851834955262e-06,
+      "loss": 0.1076,
+      "step": 3870
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.259690788014085e-06,
+      "loss": 0.0953,
+      "step": 3871
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2535015563692932e-06,
+      "loss": 0.1305,
+      "step": 3872
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2473174912834044e-06,
+      "loss": 0.1012,
+      "step": 3873
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.24113859547642e-06,
+      "loss": 0.0874,
+      "step": 3874
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2349648716660386e-06,
+      "loss": 0.0783,
+      "step": 3875
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.228796322567711e-06,
+      "loss": 0.101,
+      "step": 3876
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2226329508945982e-06,
+      "loss": 0.0733,
+      "step": 3877
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.216474759357586e-06,
+      "loss": 0.0688,
+      "step": 3878
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2103217506652803e-06,
+      "loss": 0.0874,
+      "step": 3879
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2041739275240133e-06,
+      "loss": 0.0999,
+      "step": 3880
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.198031292637827e-06,
+      "loss": 0.1019,
+      "step": 3881
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.191893848708493e-06,
+      "loss": 0.1029,
+      "step": 3882
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1857615984354926e-06,
+      "loss": 0.1141,
+      "step": 3883
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1796345445160213e-06,
+      "loss": 0.0557,
+      "step": 3884
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1735126896449935e-06,
+      "loss": 0.1184,
+      "step": 3885
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1673960365150333e-06,
+      "loss": 0.1151,
+      "step": 3886
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.161284587816476e-06,
+      "loss": 0.086,
+      "step": 3887
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.155178346237382e-06,
+      "loss": 0.0792,
+      "step": 3888
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1490773144634928e-06,
+      "loss": 0.1178,
+      "step": 3889
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1429814951782887e-06,
+      "loss": 0.08,
+      "step": 3890
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.136890891062939e-06,
+      "loss": 0.0895,
+      "step": 3891
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.130805504796319e-06,
+      "loss": 0.105,
+      "step": 3892
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1247253390550246e-06,
+      "loss": 0.0806,
+      "step": 3893
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1186503965133396e-06,
+      "loss": 0.0919,
+      "step": 3894
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.1125806798432547e-06,
+      "loss": 0.0845,
+      "step": 3895
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.106516191714467e-06,
+      "loss": 0.0869,
+      "step": 3896
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.1004569347943666e-06,
+      "loss": 0.1027,
+      "step": 3897
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0944029117480445e-06,
+      "loss": 0.0758,
+      "step": 3898
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.088354125238301e-06,
+      "loss": 0.0904,
+      "step": 3899
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0823105779256095e-06,
+      "loss": 0.0933,
+      "step": 3900
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0762722724681643e-06,
+      "loss": 0.1093,
+      "step": 3901
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.070239211521839e-06,
+      "loss": 0.1065,
+      "step": 3902
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.064211397740205e-06,
+      "loss": 0.1089,
+      "step": 3903
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0581888337745234e-06,
+      "loss": 0.0736,
+      "step": 3904
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0521715222737513e-06,
+      "loss": 0.0337,
+      "step": 3905
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0461594658845247e-06,
+      "loss": 0.0843,
+      "step": 3906
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0401526672511843e-06,
+      "loss": 0.0891,
+      "step": 3907
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.034151129015749e-06,
+      "loss": 0.0777,
+      "step": 3908
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.02815485381792e-06,
+      "loss": 0.1125,
+      "step": 3909
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0221638442950916e-06,
+      "loss": 0.0605,
+      "step": 3910
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0161781030823366e-06,
+      "loss": 0.102,
+      "step": 3911
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.010197632812412e-06,
+      "loss": 0.1223,
+      "step": 3912
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0042224361157657e-06,
+      "loss": 0.0905,
+      "step": 3913
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.9982525156205016e-06,
+      "loss": 0.1087,
+      "step": 3914
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.9922878739524327e-06,
+      "loss": 0.0754,
+      "step": 3915
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.986328513735033e-06,
+      "loss": 0.1049,
+      "step": 3916
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.9803744375894543e-06,
+      "loss": 0.0746,
+      "step": 3917
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.9744256481345276e-06,
+      "loss": 0.0935,
+      "step": 3918
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.968482147986755e-06,
+      "loss": 0.0893,
+      "step": 3919
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9625439397603237e-06,
+      "loss": 0.1319,
+      "step": 3920
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9566110260670776e-06,
+      "loss": 0.0561,
+      "step": 3921
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.950683409516541e-06,
+      "loss": 0.0876,
+      "step": 3922
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9447610927159013e-06,
+      "loss": 0.1076,
+      "step": 3923
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.938844078270034e-06,
+      "loss": 0.0971,
+      "step": 3924
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.932932368781448e-06,
+      "loss": 0.0927,
+      "step": 3925
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9270259668503542e-06,
+      "loss": 0.0912,
+      "step": 3926
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9211248750746088e-06,
+      "loss": 0.0935,
+      "step": 3927
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.915229096049738e-06,
+      "loss": 0.1092,
+      "step": 3928
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.90933863236893e-06,
+      "loss": 0.0829,
+      "step": 3929
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9034534866230366e-06,
+      "loss": 0.1027,
+      "step": 3930
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.897573661400565e-06,
+      "loss": 0.1262,
+      "step": 3931
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8916991592876962e-06,
+      "loss": 0.1554,
+      "step": 3932
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8858299828682555e-06,
+      "loss": 0.1011,
+      "step": 3933
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8799661347237326e-06,
+      "loss": 0.0919,
+      "step": 3934
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8741076174332698e-06,
+      "loss": 0.1068,
+      "step": 3935
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8682544335736686e-06,
+      "loss": 0.0934,
+      "step": 3936
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.862406585719377e-06,
+      "loss": 0.0915,
+      "step": 3937
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.856564076442517e-06,
+      "loss": 0.1396,
+      "step": 3938
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8507269083128286e-06,
+      "loss": 0.0819,
+      "step": 3939
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8448950838977356e-06,
+      "loss": 0.1054,
+      "step": 3940
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8390686057622915e-06,
+      "loss": 0.1174,
+      "step": 3941
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.833247476469205e-06,
+      "loss": 0.1155,
+      "step": 3942
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8274316985788312e-06,
+      "loss": 0.1081,
+      "step": 3943
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.8216212746491704e-06,
+      "loss": 0.1067,
+      "step": 3944
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.8158162072358667e-06,
+      "loss": 0.0756,
+      "step": 3945
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.810016498892216e-06,
+      "loss": 0.0869,
+      "step": 3946
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.8042221521691513e-06,
+      "loss": 0.0918,
+      "step": 3947
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7984331696152455e-06,
+      "loss": 0.0666,
+      "step": 3948
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.792649553776715e-06,
+      "loss": 0.0777,
+      "step": 3949
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.78687130719741e-06,
+      "loss": 0.092,
+      "step": 3950
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7810984324188343e-06,
+      "loss": 0.1257,
+      "step": 3951
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.775330931980112e-06,
+      "loss": 0.0975,
+      "step": 3952
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7695688084180134e-06,
+      "loss": 0.0844,
+      "step": 3953
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.763812064266938e-06,
+      "loss": 0.0716,
+      "step": 3954
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.758060702058922e-06,
+      "loss": 0.065,
+      "step": 3955
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7523147243236326e-06,
+      "loss": 0.1002,
+      "step": 3956
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.746574133588379e-06,
+      "loss": 0.0961,
+      "step": 3957
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7408389323780798e-06,
+      "loss": 0.1028,
+      "step": 3958
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7351091232153056e-06,
+      "loss": 0.1388,
+      "step": 3959
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.729384708620243e-06,
+      "loss": 0.1054,
+      "step": 3960
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.723665691110705e-06,
+      "loss": 0.1373,
+      "step": 3961
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7179520732021353e-06,
+      "loss": 0.0979,
+      "step": 3962
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.712243857407608e-06,
+      "loss": 0.1167,
+      "step": 3963
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7065410462378022e-06,
+      "loss": 0.064,
+      "step": 3964
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7008436422010426e-06,
+      "loss": 0.1066,
+      "step": 3965
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.6951516478032613e-06,
+      "loss": 0.1122,
+      "step": 3966
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.6894650655480155e-06,
+      "loss": 0.1169,
+      "step": 3967
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.6837838979364804e-06,
+      "loss": 0.0704,
+      "step": 3968
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6781081474674528e-06,
+      "loss": 0.0914,
+      "step": 3969
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.672437816637338e-06,
+      "loss": 0.0558,
+      "step": 3970
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.666772907940174e-06,
+      "loss": 0.0935,
+      "step": 3971
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6611134238676e-06,
+      "loss": 0.0809,
+      "step": 3972
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6554593669088723e-06,
+      "loss": 0.1161,
+      "step": 3973
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.649810739550862e-06,
+      "loss": 0.0847,
+      "step": 3974
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.64416754427805e-06,
+      "loss": 0.1001,
+      "step": 3975
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6385297835725273e-06,
+      "loss": 0.081,
+      "step": 3976
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.632897459914008e-06,
+      "loss": 0.1143,
+      "step": 3977
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6272705757797856e-06,
+      "loss": 0.0705,
+      "step": 3978
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6216491336447927e-06,
+      "loss": 0.1018,
+      "step": 3979
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6160331359815503e-06,
+      "loss": 0.1124,
+      "step": 3980
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6104225852601836e-06,
+      "loss": 0.1087,
+      "step": 3981
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.604817483948439e-06,
+      "loss": 0.083,
+      "step": 3982
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5992178345116407e-06,
+      "loss": 0.1568,
+      "step": 3983
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.593623639412737e-06,
+      "loss": 0.0947,
+      "step": 3984
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.588034901112268e-06,
+      "loss": 0.1255,
+      "step": 3985
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5824516220683715e-06,
+      "loss": 0.1001,
+      "step": 3986
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.576873804736785e-06,
+      "loss": 0.0828,
+      "step": 3987
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.571301451570857e-06,
+      "loss": 0.0885,
+      "step": 3988
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5657345650215068e-06,
+      "loss": 0.0521,
+      "step": 3989
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5601731475372723e-06,
+      "loss": 0.0789,
+      "step": 3990
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.554617201564277e-06,
+      "loss": 0.0831,
+      "step": 3991
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5490667295462378e-06,
+      "loss": 0.0924,
+      "step": 3992
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5435217339244613e-06,
+      "loss": 0.1349,
+      "step": 3993
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5379822171378524e-06,
+      "loss": 0.1063,
+      "step": 3994
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.532448181622897e-06,
+      "loss": 0.0721,
+      "step": 3995
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5269196298136855e-06,
+      "loss": 0.0697,
+      "step": 3996
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5213965641418716e-06,
+      "loss": 0.0687,
+      "step": 3997
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5158789870367227e-06,
+      "loss": 0.112,
+      "step": 3998
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.510366900925077e-06,
+      "loss": 0.0973,
+      "step": 3999
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.50486030823136e-06,
+      "loss": 0.1078,
+      "step": 4000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4993592113775778e-06,
+      "loss": 0.1336,
+      "step": 4001
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.493863612783333e-06,
+      "loss": 0.1154,
+      "step": 4002
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.488373514865786e-06,
+      "loss": 0.1135,
+      "step": 4003
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.482888920039703e-06,
+      "loss": 0.1005,
+      "step": 4004
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.477409830717415e-06,
+      "loss": 0.0918,
+      "step": 4005
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.471936249308835e-06,
+      "loss": 0.1133,
+      "step": 4006
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.466468178221451e-06,
+      "loss": 0.1442,
+      "step": 4007
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4610056198603326e-06,
+      "loss": 0.0713,
+      "step": 4008
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.455548576628116e-06,
+      "loss": 0.1089,
+      "step": 4009
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4500970509250266e-06,
+      "loss": 0.0976,
+      "step": 4010
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4446510451488472e-06,
+      "loss": 0.125,
+      "step": 4011
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4392105616949397e-06,
+      "loss": 0.1288,
+      "step": 4012
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4337756029562436e-06,
+      "loss": 0.0798,
+      "step": 4013
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4283461713232472e-06,
+      "loss": 0.123,
+      "step": 4014
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4229222691840356e-06,
+      "loss": 0.1235,
+      "step": 4015
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.417503898924244e-06,
+      "loss": 0.0581,
+      "step": 4016
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4120910629270764e-06,
+      "loss": 0.1128,
+      "step": 4017
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.406683763573308e-06,
+      "loss": 0.1199,
+      "step": 4018
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.4012820032412725e-06,
+      "loss": 0.087,
+      "step": 4019
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3958857843068693e-06,
+      "loss": 0.079,
+      "step": 4020
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.390495109143572e-06,
+      "loss": 0.117,
+      "step": 4021
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3851099801223912e-06,
+      "loss": 0.076,
+      "step": 4022
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3797303996119217e-06,
+      "loss": 0.089,
+      "step": 4023
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.374356369978307e-06,
+      "loss": 0.073,
+      "step": 4024
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.36898789358525e-06,
+      "loss": 0.07,
+      "step": 4025
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.363624972794008e-06,
+      "loss": 0.0822,
+      "step": 4026
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3582676099634094e-06,
+      "loss": 0.0944,
+      "step": 4027
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3529158074498124e-06,
+      "loss": 0.1074,
+      "step": 4028
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.347569567607156e-06,
+      "loss": 0.0953,
+      "step": 4029
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.342228892786916e-06,
+      "loss": 0.0792,
+      "step": 4030
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.336893785338125e-06,
+      "loss": 0.0685,
+      "step": 4031
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.331564247607369e-06,
+      "loss": 0.1061,
+      "step": 4032
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.326240281938781e-06,
+      "loss": 0.0712,
+      "step": 4033
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3209218906740422e-06,
+      "loss": 0.1095,
+      "step": 4034
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3156090761523933e-06,
+      "loss": 0.0688,
+      "step": 4035
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3103018407106014e-06,
+      "loss": 0.1092,
+      "step": 4036
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3050001866829982e-06,
+      "loss": 0.0986,
+      "step": 4037
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.299704116401454e-06,
+      "loss": 0.06,
+      "step": 4038
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2944136321953808e-06,
+      "loss": 0.1048,
+      "step": 4039
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.289128736391733e-06,
+      "loss": 0.0905,
+      "step": 4040
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.283849431315015e-06,
+      "loss": 0.091,
+      "step": 4041
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2785757192872676e-06,
+      "loss": 0.0949,
+      "step": 4042
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.273307602628067e-06,
+      "loss": 0.1275,
+      "step": 4043
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.268045083654533e-06,
+      "loss": 0.0716,
+      "step": 4044
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.262788164681319e-06,
+      "loss": 0.064,
+      "step": 4045
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.257536848020631e-06,
+      "loss": 0.1462,
+      "step": 4046
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.252291135982184e-06,
+      "loss": 0.05,
+      "step": 4047
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.247051030873253e-06,
+      "loss": 0.1247,
+      "step": 4048
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.241816534998632e-06,
+      "loss": 0.0697,
+      "step": 4049
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.236587650660653e-06,
+      "loss": 0.1134,
+      "step": 4050
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2313643801591747e-06,
+      "loss": 0.0661,
+      "step": 4051
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2261467257916045e-06,
+      "loss": 0.1068,
+      "step": 4052
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2209346898528484e-06,
+      "loss": 0.1224,
+      "step": 4053
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2157282746353735e-06,
+      "loss": 0.1015,
+      "step": 4054
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.210527482429152e-06,
+      "loss": 0.0738,
+      "step": 4055
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.205332315521696e-06,
+      "loss": 0.1461,
+      "step": 4056
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.200142776198035e-06,
+      "loss": 0.0794,
+      "step": 4057
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1949588667407276e-06,
+      "loss": 0.0907,
+      "step": 4058
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.189780589429851e-06,
+      "loss": 0.0917,
+      "step": 4059
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.184607946543022e-06,
+      "loss": 0.0647,
+      "step": 4060
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1794409403553516e-06,
+      "loss": 0.0903,
+      "step": 4061
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.174279573139498e-06,
+      "loss": 0.1181,
+      "step": 4062
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1691238471656207e-06,
+      "loss": 0.1009,
+      "step": 4063
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1639737647014094e-06,
+      "loss": 0.0893,
+      "step": 4064
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1588293280120614e-06,
+      "loss": 0.1018,
+      "step": 4065
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1536905393603074e-06,
+      "loss": 0.0656,
+      "step": 4066
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.148557401006368e-06,
+      "loss": 0.1002,
+      "step": 4067
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1434299152080074e-06,
+      "loss": 0.1084,
+      "step": 4068
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.138308084220481e-06,
+      "loss": 0.1032,
+      "step": 4069
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1331919102965673e-06,
+      "loss": 0.0766,
+      "step": 4070
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1280813956865646e-06,
+      "loss": 0.068,
+      "step": 4071
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1229765426382563e-06,
+      "loss": 0.1075,
+      "step": 4072
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1178773533969633e-06,
+      "loss": 0.0843,
+      "step": 4073
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1127838302055024e-06,
+      "loss": 0.1108,
+      "step": 4074
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1076959753041976e-06,
+      "loss": 0.0671,
+      "step": 4075
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.102613790930878e-06,
+      "loss": 0.0938,
+      "step": 4076
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0975372793208943e-06,
+      "loss": 0.0597,
+      "step": 4077
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.092466442707075e-06,
+      "loss": 0.0816,
+      "step": 4078
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0874012833197814e-06,
+      "loss": 0.1186,
+      "step": 4079
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.082341803386856e-06,
+      "loss": 0.1242,
+      "step": 4080
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0772880051336563e-06,
+      "loss": 0.1163,
+      "step": 4081
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.072239890783033e-06,
+      "loss": 0.1119,
+      "step": 4082
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.067197462555339e-06,
+      "loss": 0.0804,
+      "step": 4083
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0621607226684254e-06,
+      "loss": 0.0994,
+      "step": 4084
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.057129673337653e-06,
+      "loss": 0.1174,
+      "step": 4085
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0521043167758544e-06,
+      "loss": 0.0941,
+      "step": 4086
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0470846551933863e-06,
+      "loss": 0.1085,
+      "step": 4087
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0420706907980817e-06,
+      "loss": 0.0667,
+      "step": 4088
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0370624257952753e-06,
+      "loss": 0.0807,
+      "step": 4089
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0320598623877876e-06,
+      "loss": 0.0962,
+      "step": 4090
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.027063002775951e-06,
+      "loss": 0.0873,
+      "step": 4091
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0220718491575608e-06,
+      "loss": 0.0981,
+      "step": 4092
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0170864037279236e-06,
+      "loss": 0.0845,
+      "step": 4093
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0121066686798306e-06,
+      "loss": 0.1493,
+      "step": 4094
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0071326462035567e-06,
+      "loss": 0.1024,
+      "step": 4095
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0021643384868666e-06,
+      "loss": 0.0908,
+      "step": 4096
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9972017477150138e-06,
+      "loss": 0.0873,
+      "step": 4097
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9922448760707314e-06,
+      "loss": 0.0461,
+      "step": 4098
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9872937257342493e-06,
+      "loss": 0.0821,
+      "step": 4099
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9823482988832615e-06,
+      "loss": 0.0984,
+      "step": 4100
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9774085976929624e-06,
+      "loss": 0.073,
+      "step": 4101
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.972474624336026e-06,
+      "loss": 0.0969,
+      "step": 4102
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.967546380982591e-06,
+      "loss": 0.1003,
+      "step": 4103
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.962623869800296e-06,
+      "loss": 0.0728,
+      "step": 4104
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9577070929542464e-06,
+      "loss": 0.0762,
+      "step": 4105
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9527960526070294e-06,
+      "loss": 0.0911,
+      "step": 4106
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9478907509187073e-06,
+      "loss": 0.0703,
+      "step": 4107
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9429911900468188e-06,
+      "loss": 0.0772,
+      "step": 4108
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9380973721463757e-06,
+      "loss": 0.0872,
+      "step": 4109
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.933209299369876e-06,
+      "loss": 0.0847,
+      "step": 4110
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9283269738672674e-06,
+      "loss": 0.1124,
+      "step": 4111
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9234503977859935e-06,
+      "loss": 0.0778,
+      "step": 4112
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9185795732709545e-06,
+      "loss": 0.0648,
+      "step": 4113
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9137145024645263e-06,
+      "loss": 0.0863,
+      "step": 4114
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.908855187506548e-06,
+      "loss": 0.0824,
+      "step": 4115
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9040016305343445e-06,
+      "loss": 0.1098,
+      "step": 4116
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8991538336826796e-06,
+      "loss": 0.0799,
+      "step": 4117
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8943117990838132e-06,
+      "loss": 0.0905,
+      "step": 4118
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8894755288674499e-06,
+      "loss": 0.1208,
+      "step": 4119
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8846450251607678e-06,
+      "loss": 0.1026,
+      "step": 4120
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8798202900884077e-06,
+      "loss": 0.0811,
+      "step": 4121
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.875001325772472e-06,
+      "loss": 0.0962,
+      "step": 4122
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8701881343325217e-06,
+      "loss": 0.1053,
+      "step": 4123
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.865380717885592e-06,
+      "loss": 0.0957,
+      "step": 4124
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8605790785461562e-06,
+      "loss": 0.0876,
+      "step": 4125
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8557832184261685e-06,
+      "loss": 0.0639,
+      "step": 4126
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8509931396350267e-06,
+      "loss": 0.1098,
+      "step": 4127
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8462088442795915e-06,
+      "loss": 0.1189,
+      "step": 4128
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.841430334464177e-06,
+      "loss": 0.0719,
+      "step": 4129
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8366576122905582e-06,
+      "loss": 0.0932,
+      "step": 4130
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8318906798579594e-06,
+      "loss": 0.0764,
+      "step": 4131
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.82712953926306e-06,
+      "loss": 0.0976,
+      "step": 4132
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8223741925999883e-06,
+      "loss": 0.1152,
+      "step": 4133
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8176246419603292e-06,
+      "loss": 0.1197,
+      "step": 4134
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8128808894331223e-06,
+      "loss": 0.0962,
+      "step": 4135
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8081429371048397e-06,
+      "loss": 0.083,
+      "step": 4136
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.803410787059423e-06,
+      "loss": 0.1042,
+      "step": 4137
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.7986844413782504e-06,
+      "loss": 0.0976,
+      "step": 4138
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.7939639021401495e-06,
+      "loss": 0.0668,
+      "step": 4139
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7892491714213903e-06,
+      "loss": 0.084,
+      "step": 4140
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7845402512956999e-06,
+      "loss": 0.1373,
+      "step": 4141
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.779837143834231e-06,
+      "loss": 0.0563,
+      "step": 4142
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7751398511055973e-06,
+      "loss": 0.0982,
+      "step": 4143
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7704483751758443e-06,
+      "loss": 0.0929,
+      "step": 4144
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.765762718108465e-06,
+      "loss": 0.0794,
+      "step": 4145
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7610828819643877e-06,
+      "loss": 0.0821,
+      "step": 4146
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7564088688019852e-06,
+      "loss": 0.0933,
+      "step": 4147
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7517406806770619e-06,
+      "loss": 0.0672,
+      "step": 4148
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7470783196428768e-06,
+      "loss": 0.0879,
+      "step": 4149
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7424217877500997e-06,
+      "loss": 0.0835,
+      "step": 4150
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7377710870468637e-06,
+      "loss": 0.0721,
+      "step": 4151
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7331262195787195e-06,
+      "loss": 0.0746,
+      "step": 4152
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7284871873886577e-06,
+      "loss": 0.1051,
+      "step": 4153
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7238539925170976e-06,
+      "loss": 0.121,
+      "step": 4154
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7192266370019077e-06,
+      "loss": 0.0871,
+      "step": 4155
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7146051228783598e-06,
+      "loss": 0.0962,
+      "step": 4156
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7099894521791847e-06,
+      "loss": 0.0857,
+      "step": 4157
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7053796269345268e-06,
+      "loss": 0.0691,
+      "step": 4158
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7007756491719623e-06,
+      "loss": 0.0772,
+      "step": 4159
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.696177520916498e-06,
+      "loss": 0.115,
+      "step": 4160
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6915852441905615e-06,
+      "loss": 0.1014,
+      "step": 4161
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6869988210140203e-06,
+      "loss": 0.0663,
+      "step": 4162
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6824182534041537e-06,
+      "loss": 0.0958,
+      "step": 4163
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6778435433756707e-06,
+      "loss": 0.0977,
+      "step": 4164
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6732746929406994e-06,
+      "loss": 0.081,
+      "step": 4165
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6687117041088062e-06,
+      "loss": 0.1186,
+      "step": 4166
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6641545788869556e-06,
+      "loss": 0.1017,
+      "step": 4167
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6596033192795522e-06,
+      "loss": 0.0987,
+      "step": 4168
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6550579272884137e-06,
+      "loss": 0.0902,
+      "step": 4169
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6505184049127754e-06,
+      "loss": 0.0854,
+      "step": 4170
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6459847541492933e-06,
+      "loss": 0.1007,
+      "step": 4171
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6414569769920413e-06,
+      "loss": 0.0887,
+      "step": 4172
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6369350754325025e-06,
+      "loss": 0.0958,
+      "step": 4173
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6324190514595944e-06,
+      "loss": 0.1307,
+      "step": 4174
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.627908907059625e-06,
+      "loss": 0.1012,
+      "step": 4175
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6234046442163336e-06,
+      "loss": 0.0741,
+      "step": 4176
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6189062649108676e-06,
+      "loss": 0.1184,
+      "step": 4177
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6144137711217843e-06,
+      "loss": 0.1028,
+      "step": 4178
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.609927164825053e-06,
+      "loss": 0.0768,
+      "step": 4179
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6054464479940617e-06,
+      "loss": 0.1006,
+      "step": 4180
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.600971622599589e-06,
+      "loss": 0.0408,
+      "step": 4181
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.596502690609844e-06,
+      "loss": 0.0933,
+      "step": 4182
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5920396539904326e-06,
+      "loss": 0.1033,
+      "step": 4183
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5875825147043644e-06,
+      "loss": 0.0957,
+      "step": 4184
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.583131274712063e-06,
+      "loss": 0.0781,
+      "step": 4185
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5786859359713523e-06,
+      "loss": 0.1042,
+      "step": 4186
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5742465004374584e-06,
+      "loss": 0.0693,
+      "step": 4187
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5698129700630275e-06,
+      "loss": 0.0859,
+      "step": 4188
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.56538534679808e-06,
+      "loss": 0.0786,
+      "step": 4189
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5609636325900628e-06,
+      "loss": 0.0665,
+      "step": 4190
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5565478293838137e-06,
+      "loss": 0.1125,
+      "step": 4191
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.552137939121569e-06,
+      "loss": 0.0641,
+      "step": 4192
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5477339637429706e-06,
+      "loss": 0.0933,
+      "step": 4193
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5433359051850553e-06,
+      "loss": 0.084,
+      "step": 4194
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.538943765382253e-06,
+      "loss": 0.0884,
+      "step": 4195
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5345575462663997e-06,
+      "loss": 0.042,
+      "step": 4196
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5301772497667172e-06,
+      "loss": 0.1092,
+      "step": 4197
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5258028778098277e-06,
+      "loss": 0.0738,
+      "step": 4198
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5214344323197555e-06,
+      "loss": 0.1296,
+      "step": 4199
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5170719152178968e-06,
+      "loss": 0.1552,
+      "step": 4200
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5127153284230638e-06,
+      "loss": 0.1084,
+      "step": 4201
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5083646738514467e-06,
+      "loss": 0.1009,
+      "step": 4202
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5040199534166265e-06,
+      "loss": 0.0666,
+      "step": 4203
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4996811690295797e-06,
+      "loss": 0.0998,
+      "step": 4204
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4953483225986747e-06,
+      "loss": 0.1042,
+      "step": 4205
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4910214160296504e-06,
+      "loss": 0.128,
+      "step": 4206
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4867004512256578e-06,
+      "loss": 0.089,
+      "step": 4207
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.482385430087218e-06,
+      "loss": 0.0796,
+      "step": 4208
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.47807635451224e-06,
+      "loss": 0.1125,
+      "step": 4209
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4737732263960225e-06,
+      "loss": 0.1224,
+      "step": 4210
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4694760476312463e-06,
+      "loss": 0.1124,
+      "step": 4211
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4651848201079698e-06,
+      "loss": 0.0755,
+      "step": 4212
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4608995457136481e-06,
+      "loss": 0.0837,
+      "step": 4213
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4566202263330968e-06,
+      "loss": 0.1028,
+      "step": 4214
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4523468638485338e-06,
+      "loss": 0.0852,
+      "step": 4215
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4480794601395436e-06,
+      "loss": 0.1098,
+      "step": 4216
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.443818017083095e-06,
+      "loss": 0.0728,
+      "step": 4217
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.439562536553527e-06,
+      "loss": 0.0955,
+      "step": 4218
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4353130204225757e-06,
+      "loss": 0.081,
+      "step": 4219
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4310694705593259e-06,
+      "loss": 0.093,
+      "step": 4220
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4268318888302629e-06,
+      "loss": 0.1004,
+      "step": 4221
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4226002770992324e-06,
+      "loss": 0.0991,
+      "step": 4222
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.418374637227459e-06,
+      "loss": 0.0913,
+      "step": 4223
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.414154971073548e-06,
+      "loss": 0.1021,
+      "step": 4224
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4099412804934592e-06,
+      "loss": 0.0859,
+      "step": 4225
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4057335673405403e-06,
+      "loss": 0.1493,
+      "step": 4226
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4015318334655058e-06,
+      "loss": 0.082,
+      "step": 4227
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3973360807164342e-06,
+      "loss": 0.1096,
+      "step": 4228
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3931463109387792e-06,
+      "loss": 0.0833,
+      "step": 4229
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3889625259753685e-06,
+      "loss": 0.0809,
+      "step": 4230
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3847847276663782e-06,
+      "loss": 0.117,
+      "step": 4231
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3806129178493736e-06,
+      "loss": 0.0937,
+      "step": 4232
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.376447098359272e-06,
+      "loss": 0.0839,
+      "step": 4233
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3722872710283606e-06,
+      "loss": 0.08,
+      "step": 4234
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3681334376862931e-06,
+      "loss": 0.0952,
+      "step": 4235
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3639856001600793e-06,
+      "loss": 0.0859,
+      "step": 4236
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3598437602740965e-06,
+      "loss": 0.1121,
+      "step": 4237
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3557079198500927e-06,
+      "loss": 0.0911,
+      "step": 4238
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3515780807071575e-06,
+      "loss": 0.1004,
+      "step": 4239
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3474542446617583e-06,
+      "loss": 0.1092,
+      "step": 4240
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3433364135277155e-06,
+      "loss": 0.0699,
+      "step": 4241
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3392245891162087e-06,
+      "loss": 0.0708,
+      "step": 4242
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3351187732357695e-06,
+      "loss": 0.1012,
+      "step": 4243
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3310189676923051e-06,
+      "loss": 0.0848,
+      "step": 4244
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3269251742890515e-06,
+      "loss": 0.0792,
+      "step": 4245
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3228373948266264e-06,
+      "loss": 0.1067,
+      "step": 4246
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3187556311029885e-06,
+      "loss": 0.0878,
+      "step": 4247
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3146798849134517e-06,
+      "loss": 0.0609,
+      "step": 4248
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3106101580506886e-06,
+      "loss": 0.1075,
+      "step": 4249
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3065464523047155e-06,
+      "loss": 0.0632,
+      "step": 4250
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3024887694629078e-06,
+      "loss": 0.1278,
+      "step": 4251
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2984371113099914e-06,
+      "loss": 0.089,
+      "step": 4252
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2943914796280387e-06,
+      "loss": 0.0913,
+      "step": 4253
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.29035187619647e-06,
+      "loss": 0.0773,
+      "step": 4254
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2863183027920673e-06,
+      "loss": 0.1094,
+      "step": 4255
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.282290761188938e-06,
+      "loss": 0.1206,
+      "step": 4256
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2782692531585566e-06,
+      "loss": 0.1172,
+      "step": 4257
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2742537804697334e-06,
+      "loss": 0.088,
+      "step": 4258
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2702443448886274e-06,
+      "loss": 0.1038,
+      "step": 4259
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2662409481787412e-06,
+      "loss": 0.1039,
+      "step": 4260
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2622435921009228e-06,
+      "loss": 0.0737,
+      "step": 4261
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2582522784133576e-06,
+      "loss": 0.1016,
+      "step": 4262
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2542670088715863e-06,
+      "loss": 0.1208,
+      "step": 4263
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2502877852284733e-06,
+      "loss": 0.0927,
+      "step": 4264
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2463146092342404e-06,
+      "loss": 0.0987,
+      "step": 4265
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2423474826364394e-06,
+      "loss": 0.0891,
+      "step": 4266
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2383864071799634e-06,
+      "loss": 0.1075,
+      "step": 4267
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.234431384607045e-06,
+      "loss": 0.0833,
+      "step": 4268
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2304824166572593e-06,
+      "loss": 0.1237,
+      "step": 4269
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2265395050675037e-06,
+      "loss": 0.1068,
+      "step": 4270
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2226026515720297e-06,
+      "loss": 0.0907,
+      "step": 4271
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2186718579024132e-06,
+      "loss": 0.089,
+      "step": 4272
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2147471257875665e-06,
+      "loss": 0.0834,
+      "step": 4273
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2108284569537388e-06,
+      "loss": 0.0864,
+      "step": 4274
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2069158531245078e-06,
+      "loss": 0.0651,
+      "step": 4275
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2030093160207856e-06,
+      "loss": 0.085,
+      "step": 4276
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1991088473608226e-06,
+      "loss": 0.0858,
+      "step": 4277
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1952144488601845e-06,
+      "loss": 0.1029,
+      "step": 4278
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1913261222317868e-06,
+      "loss": 0.0756,
+      "step": 4279
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1874438691858575e-06,
+      "loss": 0.0846,
+      "step": 4280
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1835676914299592e-06,
+      "loss": 0.093,
+      "step": 4281
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1796975906689867e-06,
+      "loss": 0.1001,
+      "step": 4282
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.175833568605158e-06,
+      "loss": 0.115,
+      "step": 4283
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1719756269380173e-06,
+      "loss": 0.0661,
+      "step": 4284
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1681237673644334e-06,
+      "loss": 0.0878,
+      "step": 4285
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1642779915786018e-06,
+      "loss": 0.1061,
+      "step": 4286
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1604383012720387e-06,
+      "loss": 0.1115,
+      "step": 4287
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1566046981335953e-06,
+      "loss": 0.1029,
+      "step": 4288
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1527771838494262e-06,
+      "loss": 0.0823,
+      "step": 4289
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1489557601030258e-06,
+      "loss": 0.0776,
+      "step": 4290
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1451404285752003e-06,
+      "loss": 0.083,
+      "step": 4291
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1413311909440771e-06,
+      "loss": 0.0993,
+      "step": 4292
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1375280488851041e-06,
+      "loss": 0.1056,
+      "step": 4293
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1337310040710536e-06,
+      "loss": 0.1012,
+      "step": 4294
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1299400581720026e-06,
+      "loss": 0.0854,
+      "step": 4295
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1261552128553608e-06,
+      "loss": 0.0893,
+      "step": 4296
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1223764697858468e-06,
+      "loss": 0.0834,
+      "step": 4297
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1186038306254926e-06,
+      "loss": 0.0854,
+      "step": 4298
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.114837297033654e-06,
+      "loss": 0.085,
+      "step": 4299
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1110768706669927e-06,
+      "loss": 0.1027,
+      "step": 4300
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1073225531794856e-06,
+      "loss": 0.0792,
+      "step": 4301
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.103574346222435e-06,
+      "loss": 0.0911,
+      "step": 4302
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0998322514444325e-06,
+      "loss": 0.1322,
+      "step": 4303
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0960962704914047e-06,
+      "loss": 0.0837,
+      "step": 4304
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0923664050065752e-06,
+      "loss": 0.1147,
+      "step": 4305
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0886426566304803e-06,
+      "loss": 0.1016,
+      "step": 4306
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0849250270009648e-06,
+      "loss": 0.1064,
+      "step": 4307
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.081213517753194e-06,
+      "loss": 0.0564,
+      "step": 4308
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0775081305196188e-06,
+      "loss": 0.1299,
+      "step": 4309
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0738088669300172e-06,
+      "loss": 0.1008,
+      "step": 4310
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0701157286114676e-06,
+      "loss": 0.0911,
+      "step": 4311
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0664287171883475e-06,
+      "loss": 0.0872,
+      "step": 4312
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0627478342823543e-06,
+      "loss": 0.0825,
+      "step": 4313
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.05907308151247e-06,
+      "loss": 0.099,
+      "step": 4314
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0554044604950014e-06,
+      "loss": 0.0859,
+      "step": 4315
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.051741972843543e-06,
+      "loss": 0.1247,
+      "step": 4316
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0480856201689976e-06,
+      "loss": 0.0981,
+      "step": 4317
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0444354040795673e-06,
+      "loss": 0.0952,
+      "step": 4318
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0407913261807627e-06,
+      "loss": 0.099,
+      "step": 4319
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0371533880753803e-06,
+      "loss": 0.0927,
+      "step": 4320
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0335215913635298e-06,
+      "loss": 0.0857,
+      "step": 4321
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.029895937642612e-06,
+      "loss": 0.096,
+      "step": 4322
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0262764285073296e-06,
+      "loss": 0.094,
+      "step": 4323
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0226630655496805e-06,
+      "loss": 0.1361,
+      "step": 4324
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0190558503589586e-06,
+      "loss": 0.1424,
+      "step": 4325
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0154547845217515e-06,
+      "loss": 0.1149,
+      "step": 4326
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0118598696219572e-06,
+      "loss": 0.0602,
+      "step": 4327
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0082711072407425e-06,
+      "loss": 0.0569,
+      "step": 4328
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.004688498956593e-06,
+      "loss": 0.1196,
+      "step": 4329
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0011120463452717e-06,
+      "loss": 0.0924,
+      "step": 4330
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.975417509798401e-07,
+      "loss": 0.0934,
+      "step": 4331
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.939776144306473e-07,
+      "loss": 0.0761,
+      "step": 4332
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.904196382653457e-07,
+      "loss": 0.0831,
+      "step": 4333
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.868678240488588e-07,
+      "loss": 0.0746,
+      "step": 4334
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.833221733434183e-07,
+      "loss": 0.0666,
+      "step": 4335
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.79782687708532e-07,
+      "loss": 0.109,
+      "step": 4336
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.762493687010028e-07,
+      "loss": 0.0953,
+      "step": 4337
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.72722217874919e-07,
+      "loss": 0.0829,
+      "step": 4338
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.692012367816566e-07,
+      "loss": 0.1069,
+      "step": 4339
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.65686426969875e-07,
+      "loss": 0.1112,
+      "step": 4340
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.62177789985526e-07,
+      "loss": 0.0856,
+      "step": 4341
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.586753273718417e-07,
+      "loss": 0.1046,
+      "step": 4342
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.55179040669335e-07,
+      "loss": 0.0815,
+      "step": 4343
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.516889314158139e-07,
+      "loss": 0.0908,
+      "step": 4344
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.482050011463523e-07,
+      "loss": 0.0562,
+      "step": 4345
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.447272513933242e-07,
+      "loss": 0.0856,
+      "step": 4346
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.412556836863712e-07,
+      "loss": 0.0976,
+      "step": 4347
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.377902995524252e-07,
+      "loss": 0.0801,
+      "step": 4348
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.343311005156918e-07,
+      "loss": 0.0655,
+      "step": 4349
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.308780880976625e-07,
+      "loss": 0.0876,
+      "step": 4350
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.274312638170973e-07,
+      "loss": 0.1044,
+      "step": 4351
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.239906291900502e-07,
+      "loss": 0.0867,
+      "step": 4352
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.205561857298356e-07,
+      "loss": 0.1032,
+      "step": 4353
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.171279349470585e-07,
+      "loss": 0.0907,
+      "step": 4354
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.137058783495927e-07,
+      "loss": 0.1057,
+      "step": 4355
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.102900174425893e-07,
+      "loss": 0.1393,
+      "step": 4356
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.068803537284732e-07,
+      "loss": 0.1174,
+      "step": 4357
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.034768887069511e-07,
+      "loss": 0.1145,
+      "step": 4358
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.000796238749858e-07,
+      "loss": 0.0918,
+      "step": 4359
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.966885607268338e-07,
+      "loss": 0.0877,
+      "step": 4360
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.933037007540117e-07,
+      "loss": 0.1262,
+      "step": 4361
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.89925045445309e-07,
+      "loss": 0.0895,
+      "step": 4362
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.865525962867882e-07,
+      "loss": 0.0902,
+      "step": 4363
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.831863547617813e-07,
+      "loss": 0.0582,
+      "step": 4364
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.798263223508873e-07,
+      "loss": 0.0902,
+      "step": 4365
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.76472500531984e-07,
+      "loss": 0.1025,
+      "step": 4366
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.731248907801997e-07,
+      "loss": 0.1028,
+      "step": 4367
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.697834945679517e-07,
+      "loss": 0.07,
+      "step": 4368
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.664483133649076e-07,
+      "loss": 0.0949,
+      "step": 4369
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.631193486380107e-07,
+      "loss": 0.139,
+      "step": 4370
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.597966018514614e-07,
+      "loss": 0.0925,
+      "step": 4371
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.564800744667373e-07,
+      "loss": 0.0819,
+      "step": 4372
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.531697679425716e-07,
+      "loss": 0.0854,
+      "step": 4373
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.498656837349644e-07,
+      "loss": 0.0885,
+      "step": 4374
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.465678232971752e-07,
+      "loss": 0.1001,
+      "step": 4375
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.432761880797302e-07,
+      "loss": 0.1186,
+      "step": 4376
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.399907795304196e-07,
+      "loss": 0.0602,
+      "step": 4377
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.367115990942859e-07,
+      "loss": 0.0751,
+      "step": 4378
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.334386482136436e-07,
+      "loss": 0.0901,
+      "step": 4379
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.301719283280596e-07,
+      "loss": 0.095,
+      "step": 4380
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.269114408743611e-07,
+      "loss": 0.0944,
+      "step": 4381
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.236571872866327e-07,
+      "loss": 0.1241,
+      "step": 4382
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.204091689962279e-07,
+      "loss": 0.0897,
+      "step": 4383
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.171673874317387e-07,
+      "loss": 0.0721,
+      "step": 4384
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.139318440190313e-07,
+      "loss": 0.0661,
+      "step": 4385
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.107025401812207e-07,
+      "loss": 0.0892,
+      "step": 4386
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.074794773386773e-07,
+      "loss": 0.064,
+      "step": 4387
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.042626569090272e-07,
+      "loss": 0.0612,
+      "step": 4388
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.010520803071503e-07,
+      "loss": 0.1024,
+      "step": 4389
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.978477489451785e-07,
+      "loss": 0.0908,
+      "step": 4390
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.946496642325079e-07,
+      "loss": 0.0611,
+      "step": 4391
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.914578275757684e-07,
+      "loss": 0.1191,
+      "step": 4392
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.882722403788567e-07,
+      "loss": 0.071,
+      "step": 4393
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.850929040429156e-07,
+      "loss": 0.0781,
+      "step": 4394
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.819198199663379e-07,
+      "loss": 0.0589,
+      "step": 4395
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.78752989544766e-07,
+      "loss": 0.0822,
+      "step": 4396
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.755924141710985e-07,
+      "loss": 0.1068,
+      "step": 4397
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.724380952354704e-07,
+      "loss": 0.0789,
+      "step": 4398
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.692900341252751e-07,
+      "loss": 0.1238,
+      "step": 4399
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.661482322251484e-07,
+      "loss": 0.107,
+      "step": 4400
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.630126909169782e-07,
+      "loss": 0.1183,
+      "step": 4401
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.598834115798913e-07,
+      "loss": 0.1038,
+      "step": 4402
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.567603955902652e-07,
+      "loss": 0.0889,
+      "step": 4403
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.53643644321726e-07,
+      "loss": 0.1142,
+      "step": 4404
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.505331591451353e-07,
+      "loss": 0.0862,
+      "step": 4405
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.47428941428604e-07,
+      "loss": 0.144,
+      "step": 4406
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.443309925374848e-07,
+      "loss": 0.0842,
+      "step": 4407
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.412393138343809e-07,
+      "loss": 0.0818,
+      "step": 4408
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.381539066791198e-07,
+      "loss": 0.0979,
+      "step": 4409
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.35074772428791e-07,
+      "loss": 0.0917,
+      "step": 4410
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.320019124377108e-07,
+      "loss": 0.1147,
+      "step": 4411
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.289353280574418e-07,
+      "loss": 0.1026,
+      "step": 4412
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.258750206367831e-07,
+      "loss": 0.0607,
+      "step": 4413
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.228209915217765e-07,
+      "loss": 0.0708,
+      "step": 4414
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.197732420556991e-07,
+      "loss": 0.0671,
+      "step": 4415
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.167317735790735e-07,
+      "loss": 0.1057,
+      "step": 4416
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.13696587429643e-07,
+      "loss": 0.0784,
+      "step": 4417
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.106676849424082e-07,
+      "loss": 0.1055,
+      "step": 4418
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.076450674495932e-07,
+      "loss": 0.1195,
+      "step": 4419
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.046287362806581e-07,
+      "loss": 0.1329,
+      "step": 4420
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.016186927623014e-07,
+      "loss": 0.0975,
+      "step": 4421
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.986149382184609e-07,
+      "loss": 0.1039,
+      "step": 4422
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.956174739702947e-07,
+      "loss": 0.0859,
+      "step": 4423
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.926263013362066e-07,
+      "loss": 0.0782,
+      "step": 4424
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.896414216318275e-07,
+      "loss": 0.0887,
+      "step": 4425
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.866628361700223e-07,
+      "loss": 0.1042,
+      "step": 4426
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.836905462608861e-07,
+      "loss": 0.0666,
+      "step": 4427
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.807245532117451e-07,
+      "loss": 0.1234,
+      "step": 4428
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.777648583271557e-07,
+      "loss": 0.0949,
+      "step": 4429
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.748114629089097e-07,
+      "loss": 0.111,
+      "step": 4430
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.71864368256015e-07,
+      "loss": 0.0976,
+      "step": 4431
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.689235756647244e-07,
+      "loss": 0.0844,
+      "step": 4432
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.659890864285067e-07,
+      "loss": 0.1131,
+      "step": 4433
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.630609018380596e-07,
+      "loss": 0.0574,
+      "step": 4434
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.601390231813193e-07,
+      "loss": 0.0685,
+      "step": 4435
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.572234517434327e-07,
+      "loss": 0.1115,
+      "step": 4436
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.54314188806781e-07,
+      "loss": 0.1217,
+      "step": 4437
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.514112356509705e-07,
+      "loss": 0.0967,
+      "step": 4438
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.485145935528297e-07,
+      "loss": 0.063,
+      "step": 4439
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.456242637864113e-07,
+      "loss": 0.0588,
+      "step": 4440
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.427402476229982e-07,
+      "loss": 0.0612,
+      "step": 4441
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.398625463310825e-07,
+      "loss": 0.0917,
+      "step": 4442
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.369911611763935e-07,
+      "loss": 0.1106,
+      "step": 4443
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.341260934218757e-07,
+      "loss": 0.1033,
+      "step": 4444
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.312673443276928e-07,
+      "loss": 0.1271,
+      "step": 4445
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.284149151512325e-07,
+      "loss": 0.0791,
+      "step": 4446
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.25568807147106e-07,
+      "loss": 0.0857,
+      "step": 4447
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.227290215671339e-07,
+      "loss": 0.0966,
+      "step": 4448
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.198955596603667e-07,
+      "loss": 0.1086,
+      "step": 4449
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.170684226730694e-07,
+      "loss": 0.0653,
+      "step": 4450
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.142476118487222e-07,
+      "loss": 0.1351,
+      "step": 4451
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.114331284280278e-07,
+      "loss": 0.0746,
+      "step": 4452
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.086249736489025e-07,
+      "loss": 0.082,
+      "step": 4453
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.0582314874648e-07,
+      "loss": 0.0735,
+      "step": 4454
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.030276549531127e-07,
+      "loss": 0.1098,
+      "step": 4455
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.00238493498359e-07,
+      "loss": 0.137,
+      "step": 4456
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.974556656090058e-07,
+      "loss": 0.1032,
+      "step": 4457
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.946791725090439e-07,
+      "loss": 0.0901,
+      "step": 4458
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.919090154196815e-07,
+      "loss": 0.1224,
+      "step": 4459
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.891451955593374e-07,
+      "loss": 0.0901,
+      "step": 4460
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.863877141436525e-07,
+      "loss": 0.1029,
+      "step": 4461
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.836365723854614e-07,
+      "loss": 0.0756,
+      "step": 4462
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.808917714948298e-07,
+      "loss": 0.1045,
+      "step": 4463
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.781533126790251e-07,
+      "loss": 0.0878,
+      "step": 4464
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.754211971425205e-07,
+      "loss": 0.0964,
+      "step": 4465
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.726954260870132e-07,
+      "loss": 0.1375,
+      "step": 4466
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.699760007113941e-07,
+      "loss": 0.0751,
+      "step": 4467
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.672629222117748e-07,
+      "loss": 0.0662,
+      "step": 4468
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.645561917814707e-07,
+      "loss": 0.0812,
+      "step": 4469
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.618558106110027e-07,
+      "loss": 0.1006,
+      "step": 4470
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.591617798881026e-07,
+      "loss": 0.0727,
+      "step": 4471
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.564741007977092e-07,
+      "loss": 0.1066,
+      "step": 4472
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.53792774521965e-07,
+      "loss": 0.1174,
+      "step": 4473
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.511178022402219e-07,
+      "loss": 0.1041,
+      "step": 4474
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.484491851290335e-07,
+      "loss": 0.0806,
+      "step": 4475
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.457869243621593e-07,
+      "loss": 0.1053,
+      "step": 4476
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.431310211105628e-07,
+      "loss": 0.084,
+      "step": 4477
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.404814765424143e-07,
+      "loss": 0.0852,
+      "step": 4478
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.378382918230801e-07,
+      "loss": 0.1042,
+      "step": 4479
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.352014681151419e-07,
+      "loss": 0.0914,
+      "step": 4480
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.325710065783651e-07,
+      "loss": 0.1087,
+      "step": 4481
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.299469083697361e-07,
+      "loss": 0.0797,
+      "step": 4482
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.273291746434283e-07,
+      "loss": 0.0855,
+      "step": 4483
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.24717806550824e-07,
+      "loss": 0.1022,
+      "step": 4484
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.221128052404995e-07,
+      "loss": 0.0854,
+      "step": 4485
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.195141718582397e-07,
+      "loss": 0.1099,
+      "step": 4486
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.169219075470155e-07,
+      "loss": 0.0778,
+      "step": 4487
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.143360134470082e-07,
+      "loss": 0.0958,
+      "step": 4488
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.117564906955929e-07,
+      "loss": 0.0998,
+      "step": 4489
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.091833404273422e-07,
+      "loss": 0.0947,
+      "step": 4490
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.066165637740239e-07,
+      "loss": 0.0789,
+      "step": 4491
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.040561618646033e-07,
+      "loss": 0.0738,
+      "step": 4492
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.015021358252497e-07,
+      "loss": 0.1211,
+      "step": 4493
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.989544867793144e-07,
+      "loss": 0.1048,
+      "step": 4494
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.964132158473544e-07,
+      "loss": 0.085,
+      "step": 4495
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.938783241471173e-07,
+      "loss": 0.0829,
+      "step": 4496
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.913498127935445e-07,
+      "loss": 0.0866,
+      "step": 4497
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.888276828987697e-07,
+      "loss": 0.0862,
+      "step": 4498
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.863119355721273e-07,
+      "loss": 0.0795,
+      "step": 4499
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.838025719201372e-07,
+      "loss": 0.1338,
+      "step": 4500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.812995930465114e-07,
+      "loss": 0.105,
+      "step": 4501
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.788030000521576e-07,
+      "loss": 0.0972,
+      "step": 4502
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.763127940351708e-07,
+      "loss": 0.085,
+      "step": 4503
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.738289760908399e-07,
+      "loss": 0.0749,
+      "step": 4504
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.7135154731164586e-07,
+      "loss": 0.0837,
+      "step": 4505
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.688805087872505e-07,
+      "loss": 0.0964,
+      "step": 4506
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.664158616045161e-07,
+      "loss": 0.0907,
+      "step": 4507
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.639576068474871e-07,
+      "loss": 0.0709,
+      "step": 4508
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.61505745597397e-07,
+      "loss": 0.1297,
+      "step": 4509
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.590602789326681e-07,
+      "loss": 0.1006,
+      "step": 4510
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.566212079289117e-07,
+      "loss": 0.069,
+      "step": 4511
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.54188533658923e-07,
+      "loss": 0.1056,
+      "step": 4512
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.5176225719268607e-07,
+      "loss": 0.0725,
+      "step": 4513
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.493423795973689e-07,
+      "loss": 0.0759,
+      "step": 4514
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.4692890193732685e-07,
+      "loss": 0.0559,
+      "step": 4515
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.4452182527410067e-07,
+      "loss": 0.0839,
+      "step": 4516
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.421211506664119e-07,
+      "loss": 0.0751,
+      "step": 4517
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3972687917017087e-07,
+      "loss": 0.0898,
+      "step": 4518
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.373390118384718e-07,
+      "loss": 0.0653,
+      "step": 4519
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3495754972158464e-07,
+      "loss": 0.122,
+      "step": 4520
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.325824938669715e-07,
+      "loss": 0.1038,
+      "step": 4521
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3021384531927166e-07,
+      "loss": 0.0718,
+      "step": 4522
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.278516051203052e-07,
+      "loss": 0.0998,
+      "step": 4523
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.2549577430907927e-07,
+      "loss": 0.1049,
+      "step": 4524
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.2314635392177834e-07,
+      "loss": 0.0809,
+      "step": 4525
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.208033449917642e-07,
+      "loss": 0.0924,
+      "step": 4526
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.184667485495824e-07,
+      "loss": 0.0766,
+      "step": 4527
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.1613656562295933e-07,
+      "loss": 0.0685,
+      "step": 4528
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.13812797236795e-07,
+      "loss": 0.0838,
+      "step": 4529
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.114954444131769e-07,
+      "loss": 0.1006,
+      "step": 4530
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0918450817135965e-07,
+      "loss": 0.0866,
+      "step": 4531
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.068799895277853e-07,
+      "loss": 0.1115,
+      "step": 4532
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0458188949606787e-07,
+      "loss": 0.0795,
+      "step": 4533
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0229020908700066e-07,
+      "loss": 0.1183,
+      "step": 4534
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.000049493085489e-07,
+      "loss": 0.1056,
+      "step": 4535
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.977261111658603e-07,
+      "loss": 0.0811,
+      "step": 4536
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.954536956612531e-07,
+      "loss": 0.0927,
+      "step": 4537
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.9318770379422443e-07,
+      "loss": 0.1396,
+      "step": 4538
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.9092813656144365e-07,
+      "loss": 0.1155,
+      "step": 4539
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8867499495675416e-07,
+      "loss": 0.0758,
+      "step": 4540
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8642827997117325e-07,
+      "loss": 0.0619,
+      "step": 4541
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.841879925928937e-07,
+      "loss": 0.0828,
+      "step": 4542
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.819541338072757e-07,
+      "loss": 0.0982,
+      "step": 4543
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7972670459686334e-07,
+      "loss": 0.1023,
+      "step": 4544
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7750570594135614e-07,
+      "loss": 0.0871,
+      "step": 4545
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7529113881763944e-07,
+      "loss": 0.0989,
+      "step": 4546
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7308300419976403e-07,
+      "loss": 0.0869,
+      "step": 4547
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7088130305894975e-07,
+      "loss": 0.0653,
+      "step": 4548
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.686860363635919e-07,
+      "loss": 0.0633,
+      "step": 4549
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6649720507924976e-07,
+      "loss": 0.0459,
+      "step": 4550
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6431481016865664e-07,
+      "loss": 0.069,
+      "step": 4551
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.62138852591713e-07,
+      "loss": 0.0604,
+      "step": 4552
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.599693333054882e-07,
+      "loss": 0.0896,
+      "step": 4553
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.57806253264219e-07,
+      "loss": 0.1009,
+      "step": 4554
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.5564961341931266e-07,
+      "loss": 0.0894,
+      "step": 4555
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.534994147193388e-07,
+      "loss": 0.1073,
+      "step": 4556
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.513556581100391e-07,
+      "loss": 0.1136,
+      "step": 4557
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.492183445343194e-07,
+      "loss": 0.0959,
+      "step": 4558
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.470874749322528e-07,
+      "loss": 0.07,
+      "step": 4559
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.4496305024107446e-07,
+      "loss": 0.0959,
+      "step": 4560
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.428450713951886e-07,
+      "loss": 0.0691,
+      "step": 4561
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.4073353932616346e-07,
+      "loss": 0.0837,
+      "step": 4562
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.3862845496273274e-07,
+      "loss": 0.1073,
+      "step": 4563
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.3652981923079254e-07,
+      "loss": 0.0536,
+      "step": 4564
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.344376330534027e-07,
+      "loss": 0.0959,
+      "step": 4565
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.3235189735078555e-07,
+      "loss": 0.1056,
+      "step": 4566
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.3027261304032883e-07,
+      "loss": 0.0869,
+      "step": 4567
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.281997810365794e-07,
+      "loss": 0.1087,
+      "step": 4568
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.2613340225125134e-07,
+      "loss": 0.0726,
+      "step": 4569
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.2407347759321246e-07,
+      "loss": 0.1083,
+      "step": 4570
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.220200079685015e-07,
+      "loss": 0.1032,
+      "step": 4571
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1997299428030923e-07,
+      "loss": 0.1233,
+      "step": 4572
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1793243742899216e-07,
+      "loss": 0.0983,
+      "step": 4573
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1589833831206395e-07,
+      "loss": 0.0849,
+      "step": 4574
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.13870697824199e-07,
+      "loss": 0.0894,
+      "step": 4575
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.118495168572305e-07,
+      "loss": 0.1196,
+      "step": 4576
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.098347963001541e-07,
+      "loss": 0.0855,
+      "step": 4577
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.078265370391192e-07,
+      "loss": 0.0625,
+      "step": 4578
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.0582473995743597e-07,
+      "loss": 0.0732,
+      "step": 4579
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.0382940593556843e-07,
+      "loss": 0.0959,
+      "step": 4580
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.0184053585114444e-07,
+      "loss": 0.085,
+      "step": 4581
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9985813057894107e-07,
+      "loss": 0.0921,
+      "step": 4582
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9788219099089744e-07,
+      "loss": 0.0831,
+      "step": 4583
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9591271795610995e-07,
+      "loss": 0.0515,
+      "step": 4584
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.939497123408258e-07,
+      "loss": 0.0932,
+      "step": 4585
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9199317500844925e-07,
+      "loss": 0.0625,
+      "step": 4586
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.900431068195403e-07,
+      "loss": 0.0823,
+      "step": 4587
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.880995086318161e-07,
+      "loss": 0.0927,
+      "step": 4588
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.861623813001413e-07,
+      "loss": 0.0899,
+      "step": 4589
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8423172567654253e-07,
+      "loss": 0.0732,
+      "step": 4590
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.82307542610194e-07,
+      "loss": 0.0776,
+      "step": 4591
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.803898329474236e-07,
+      "loss": 0.0803,
+      "step": 4592
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.78478597531715e-07,
+      "loss": 0.0686,
+      "step": 4593
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.765738372037041e-07,
+      "loss": 0.1279,
+      "step": 4594
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.746755528011741e-07,
+      "loss": 0.1046,
+      "step": 4595
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.7278374515906534e-07,
+      "loss": 0.0597,
+      "step": 4596
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.7089841510946566e-07,
+      "loss": 0.102,
+      "step": 4597
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6901956348161506e-07,
+      "loss": 0.112,
+      "step": 4598
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.671471911019058e-07,
+      "loss": 0.074,
+      "step": 4599
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.652812987938791e-07,
+      "loss": 0.1252,
+      "step": 4600
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6342188737822184e-07,
+      "loss": 0.1171,
+      "step": 4601
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6156895767277975e-07,
+      "loss": 0.1201,
+      "step": 4602
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.597225104925377e-07,
+      "loss": 0.1023,
+      "step": 4603
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5788254664963764e-07,
+      "loss": 0.1167,
+      "step": 4604
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.560490669533622e-07,
+      "loss": 0.1002,
+      "step": 4605
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5422207221014793e-07,
+      "loss": 0.0945,
+      "step": 4606
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.524015632235738e-07,
+      "loss": 0.1134,
+      "step": 4607
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5058754079437587e-07,
+      "loss": 0.1228,
+      "step": 4608
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4878000572042423e-07,
+      "loss": 0.0863,
+      "step": 4609
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.469789587967447e-07,
+      "loss": 0.1017,
+      "step": 4610
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.451844008155052e-07,
+      "loss": 0.0995,
+      "step": 4611
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.433963325660227e-07,
+      "loss": 0.1077,
+      "step": 4612
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4161475483475636e-07,
+      "loss": 0.0741,
+      "step": 4613
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.398396684053128e-07,
+      "loss": 0.0643,
+      "step": 4614
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.380710740584424e-07,
+      "loss": 0.1319,
+      "step": 4615
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3630897257204132e-07,
+      "loss": 0.056,
+      "step": 4616
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3455336472114952e-07,
+      "loss": 0.1052,
+      "step": 4617
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3280425127794768e-07,
+      "loss": 0.0942,
+      "step": 4618
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.310616330117671e-07,
+      "loss": 0.0847,
+      "step": 4619
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2932551068907458e-07,
+      "loss": 0.0863,
+      "step": 4620
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2759588507348273e-07,
+      "loss": 0.1034,
+      "step": 4621
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2587275692574961e-07,
+      "loss": 0.1009,
+      "step": 4622
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.241561270037723e-07,
+      "loss": 0.0857,
+      "step": 4623
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2244599606258853e-07,
+      "loss": 0.0821,
+      "step": 4624
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2074236485437993e-07,
+      "loss": 0.1195,
+      "step": 4625
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.1904523412846712e-07,
+      "loss": 0.0707,
+      "step": 4626
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.1735460463131628e-07,
+      "loss": 0.0889,
+      "step": 4627
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.1567047710652933e-07,
+      "loss": 0.1025,
+      "step": 4628
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.1399285229485034e-07,
+      "loss": 0.0698,
+      "step": 4629
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.123217309341624e-07,
+      "loss": 0.1008,
+      "step": 4630
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.1065711375948758e-07,
+      "loss": 0.1175,
+      "step": 4631
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0899900150298855e-07,
+      "loss": 0.0957,
+      "step": 4632
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0734739489397026e-07,
+      "loss": 0.0949,
+      "step": 4633
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0570229465886493e-07,
+      "loss": 0.0921,
+      "step": 4634
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0406370152125708e-07,
+      "loss": 0.0654,
+      "step": 4635
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0243161620186024e-07,
+      "loss": 0.0888,
+      "step": 4636
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0080603941852683e-07,
+      "loss": 0.088,
+      "step": 4637
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.991869718862499e-07,
+      "loss": 0.0682,
+      "step": 4638
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9757441431715485e-07,
+      "loss": 0.1371,
+      "step": 4639
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.959683674205076e-07,
+      "loss": 0.0785,
+      "step": 4640
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9436883190270817e-07,
+      "loss": 0.1028,
+      "step": 4641
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9277580846729537e-07,
+      "loss": 0.1074,
+      "step": 4642
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.911892978149371e-07,
+      "loss": 0.1172,
+      "step": 4643
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.896093006434485e-07,
+      "loss": 0.0978,
+      "step": 4644
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8803581764776535e-07,
+      "loss": 0.0997,
+      "step": 4645
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8646884951997233e-07,
+      "loss": 0.0742,
+      "step": 4646
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8490839694927653e-07,
+      "loss": 0.0972,
+      "step": 4647
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.833544606220272e-07,
+      "loss": 0.1343,
+      "step": 4648
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8180704122170433e-07,
+      "loss": 0.0772,
+      "step": 4649
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8026613942892177e-07,
+      "loss": 0.1099,
+      "step": 4650
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7873175592142743e-07,
+      "loss": 0.0853,
+      "step": 4651
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.772038913741031e-07,
+      "loss": 0.0891,
+      "step": 4652
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.7568254645895786e-07,
+      "loss": 0.1012,
+      "step": 4653
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.7416772184514319e-07,
+      "loss": 0.0964,
+      "step": 4654
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.7265941819893272e-07,
+      "loss": 0.0768,
+      "step": 4655
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.7115763618373747e-07,
+      "loss": 0.1066,
+      "step": 4656
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6966237646009574e-07,
+      "loss": 0.0734,
+      "step": 4657
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6817363968568645e-07,
+      "loss": 0.0606,
+      "step": 4658
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6669142651530578e-07,
+      "loss": 0.0994,
+      "step": 4659
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6521573760089225e-07,
+      "loss": 0.1095,
+      "step": 4660
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6374657359150836e-07,
+      "loss": 0.0954,
+      "step": 4661
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6228393513334882e-07,
+      "loss": 0.0815,
+      "step": 4662
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6082782286973907e-07,
+      "loss": 0.094,
+      "step": 4663
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5937823744113013e-07,
+      "loss": 0.0961,
+      "step": 4664
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.579351794851086e-07,
+      "loss": 0.1279,
+      "step": 4665
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.564986496363835e-07,
+      "loss": 0.0625,
+      "step": 4666
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.550686485267977e-07,
+      "loss": 0.0785,
+      "step": 4667
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5364517678531974e-07,
+      "loss": 0.073,
+      "step": 4668
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5222823503804717e-07,
+      "loss": 0.1048,
+      "step": 4669
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5081782390820475e-07,
+      "loss": 0.0686,
+      "step": 4670
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4941394401614294e-07,
+      "loss": 0.0908,
+      "step": 4671
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4801659597934615e-07,
+      "loss": 0.097,
+      "step": 4672
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.466257804124177e-07,
+      "loss": 0.0791,
+      "step": 4673
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4524149792709328e-07,
+      "loss": 0.1257,
+      "step": 4674
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4386374913223255e-07,
+      "loss": 0.076,
+      "step": 4675
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.424925346338224e-07,
+      "loss": 0.0899,
+      "step": 4676
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4112785503497706e-07,
+      "loss": 0.081,
+      "step": 4677
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3976971093593138e-07,
+      "loss": 0.0846,
+      "step": 4678
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3841810293405088e-07,
+      "loss": 0.1056,
+      "step": 4679
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.37073031623825e-07,
+      "loss": 0.0907,
+      "step": 4680
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3573449759686552e-07,
+      "loss": 0.0945,
+      "step": 4681
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3440250144191313e-07,
+      "loss": 0.0993,
+      "step": 4682
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.330770437448292e-07,
+      "loss": 0.1119,
+      "step": 4683
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3175812508860074e-07,
+      "loss": 0.1006,
+      "step": 4684
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3044574605334037e-07,
+      "loss": 0.0873,
+      "step": 4685
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2913990721627966e-07,
+      "loss": 0.1064,
+      "step": 4686
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.278406091517792e-07,
+      "loss": 0.0951,
+      "step": 4687
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2654785243131685e-07,
+      "loss": 0.101,
+      "step": 4688
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.252616376234994e-07,
+      "loss": 0.0646,
+      "step": 4689
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2398196529405104e-07,
+      "loss": 0.0963,
+      "step": 4690
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2270883600582317e-07,
+      "loss": 0.0998,
+      "step": 4691
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2144225031878452e-07,
+      "loss": 0.0942,
+      "step": 4692
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2018220879002784e-07,
+      "loss": 0.063,
+      "step": 4693
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1892871197376808e-07,
+      "loss": 0.0945,
+      "step": 4694
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.176817604213426e-07,
+      "loss": 0.1034,
+      "step": 4695
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1644135468120432e-07,
+      "loss": 0.0941,
+      "step": 4696
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1520749529893682e-07,
+      "loss": 0.1066,
+      "step": 4697
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1398018281723266e-07,
+      "loss": 0.1045,
+      "step": 4698
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1275941777591503e-07,
+      "loss": 0.0798,
+      "step": 4699
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1154520071192109e-07,
+      "loss": 0.0865,
+      "step": 4700
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.103375321593103e-07,
+      "loss": 0.0834,
+      "step": 4701
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0913641264926111e-07,
+      "loss": 0.0762,
+      "step": 4702
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0794184271007257e-07,
+      "loss": 0.0866,
+      "step": 4703
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0675382286716106e-07,
+      "loss": 0.0772,
+      "step": 4704
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0557235364306528e-07,
+      "loss": 0.1024,
+      "step": 4705
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0439743555743952e-07,
+      "loss": 0.1356,
+      "step": 4706
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0322906912705543e-07,
+      "loss": 0.0944,
+      "step": 4707
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0206725486580859e-07,
+      "loss": 0.0913,
+      "step": 4708
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0091199328470691e-07,
+      "loss": 0.1019,
+      "step": 4709
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.976328489187891e-08,
+      "loss": 0.0891,
+      "step": 4710
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.862113019257214e-08,
+      "loss": 0.0736,
+      "step": 4711
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.748552968914803e-08,
+      "loss": 0.0878,
+      "step": 4712
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.635648388108708e-08,
+      "loss": 0.1092,
+      "step": 4713
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.523399326498872e-08,
+      "loss": 0.0696,
+      "step": 4714
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.411805833456465e-08,
+      "loss": 0.0855,
+      "step": 4715
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.300867958064896e-08,
+      "loss": 0.0783,
+      "step": 4716
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.190585749118463e-08,
+      "loss": 0.0755,
+      "step": 4717
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.080959255123866e-08,
+      "loss": 0.1264,
+      "step": 4718
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.971988524298869e-08,
+      "loss": 0.0933,
+      "step": 4719
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.863673604572964e-08,
+      "loss": 0.1298,
+      "step": 4720
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.75601454358721e-08,
+      "loss": 0.0704,
+      "step": 4721
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.649011388694228e-08,
+      "loss": 0.0649,
+      "step": 4722
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.542664186957871e-08,
+      "loss": 0.1057,
+      "step": 4723
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.43697298515389e-08,
+      "loss": 0.0847,
+      "step": 4724
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.331937829769431e-08,
+      "loss": 0.104,
+      "step": 4725
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.227558767002707e-08,
+      "loss": 0.1069,
+      "step": 4726
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.123835842763827e-08,
+      "loss": 0.0707,
+      "step": 4727
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.020769102673964e-08,
+      "loss": 0.0537,
+      "step": 4728
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.918358592065689e-08,
+      "loss": 0.0597,
+      "step": 4729
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.81660435598347e-08,
+      "loss": 0.1229,
+      "step": 4730
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.71550643918234e-08,
+      "loss": 0.0764,
+      "step": 4731
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.615064886129231e-08,
+      "loss": 0.1107,
+      "step": 4732
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.515279741002135e-08,
+      "loss": 0.0917,
+      "step": 4733
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.416151047690112e-08,
+      "loss": 0.0671,
+      "step": 4734
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.317678849794119e-08,
+      "loss": 0.0655,
+      "step": 4735
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.219863190625841e-08,
+      "loss": 0.0852,
+      "step": 4736
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.122704113208367e-08,
+      "loss": 0.0856,
+      "step": 4737
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.026201660275844e-08,
+      "loss": 0.1041,
+      "step": 4738
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.930355874273985e-08,
+      "loss": 0.0979,
+      "step": 4739
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.835166797359237e-08,
+      "loss": 0.0624,
+      "step": 4740
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.740634471399442e-08,
+      "loss": 0.1253,
+      "step": 4741
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.646758937973508e-08,
+      "loss": 0.0943,
+      "step": 4742
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.553540238371402e-08,
+      "loss": 0.1152,
+      "step": 4743
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.460978413594498e-08,
+      "loss": 0.0645,
+      "step": 4744
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.369073504355061e-08,
+      "loss": 0.0776,
+      "step": 4745
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.277825551075922e-08,
+      "loss": 0.0791,
+      "step": 4746
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.187234593891978e-08,
+      "loss": 0.0645,
+      "step": 4747
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.097300672648021e-08,
+      "loss": 0.0996,
+      "step": 4748
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.008023826901076e-08,
+      "loss": 0.0844,
+      "step": 4749
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.919404095917902e-08,
+      "loss": 0.0829,
+      "step": 4750
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8314415186773166e-08,
+      "loss": 0.0987,
+      "step": 4751
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.744136133868372e-08,
+      "loss": 0.0919,
+      "step": 4752
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.6574879798913515e-08,
+      "loss": 0.0858,
+      "step": 4753
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.57149709485727e-08,
+      "loss": 0.1059,
+      "step": 4754
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.4861635165887046e-08,
+      "loss": 0.1194,
+      "step": 4755
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.4014872826179674e-08,
+      "loss": 0.1072,
+      "step": 4756
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.317468430189432e-08,
+      "loss": 0.0948,
+      "step": 4757
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.234106996257537e-08,
+      "loss": 0.0559,
+      "step": 4758
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.1514030174877877e-08,
+      "loss": 0.074,
+      "step": 4759
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.069356530256752e-08,
+      "loss": 0.0722,
+      "step": 4760
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.987967570651564e-08,
+      "loss": 0.0703,
+      "step": 4761
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.907236174470087e-08,
+      "loss": 0.0808,
+      "step": 4762
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.827162377221084e-08,
+      "loss": 0.088,
+      "step": 4763
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.747746214124049e-08,
+      "loss": 0.0876,
+      "step": 4764
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.668987720109374e-08,
+      "loss": 0.0791,
+      "step": 4765
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.5908869298181835e-08,
+      "loss": 0.1189,
+      "step": 4766
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.513443877601831e-08,
+      "loss": 0.078,
+      "step": 4767
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.436658597522902e-08,
+      "loss": 0.1085,
+      "step": 4768
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.3605311233545495e-08,
+      "loss": 0.0743,
+      "step": 4769
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.285061488580322e-08,
+      "loss": 0.1272,
+      "step": 4770
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.210249726394999e-08,
+      "loss": 0.0709,
+      "step": 4771
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.136095869703427e-08,
+      "loss": 0.1146,
+      "step": 4772
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0625999511213486e-08,
+      "loss": 0.0973,
+      "step": 4773
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.9897620029750724e-08,
+      "loss": 0.0991,
+      "step": 4774
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.917582057301472e-08,
+      "loss": 0.1026,
+      "step": 4775
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.8460601458481515e-08,
+      "loss": 0.0895,
+      "step": 4776
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.775196300073114e-08,
+      "loss": 0.0847,
+      "step": 4777
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.704990551145093e-08,
+      "loss": 0.1068,
+      "step": 4778
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.635442929942889e-08,
+      "loss": 0.0864,
+      "step": 4779
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.5665534670566994e-08,
+      "loss": 0.097,
+      "step": 4780
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.498322192786452e-08,
+      "loss": 0.0837,
+      "step": 4781
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.430749137142974e-08,
+      "loss": 0.092,
+      "step": 4782
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.363834329847326e-08,
+      "loss": 0.0897,
+      "step": 4783
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.297577800331464e-08,
+      "loss": 0.0696,
+      "step": 4784
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.2319795777372427e-08,
+      "loss": 0.1292,
+      "step": 4785
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.1670396909175836e-08,
+      "loss": 0.0788,
+      "step": 4786
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.1027581684353044e-08,
+      "loss": 0.0739,
+      "step": 4787
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.03913503856379e-08,
+      "loss": 0.1048,
+      "step": 4788
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.9761703292871535e-08,
+      "loss": 0.0779,
+      "step": 4789
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.9138640682995764e-08,
+      "loss": 0.0752,
+      "step": 4790
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.852216283005804e-08,
+      "loss": 0.0977,
+      "step": 4791
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.791227000520813e-08,
+      "loss": 0.1003,
+      "step": 4792
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.7308962476699784e-08,
+      "loss": 0.0781,
+      "step": 4793
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.6712240509892405e-08,
+      "loss": 0.0838,
+      "step": 4794
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.6122104367246047e-08,
+      "loss": 0.0847,
+      "step": 4795
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.553855430832308e-08,
+      "loss": 0.0664,
+      "step": 4796
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4961590589796523e-08,
+      "loss": 0.0904,
+      "step": 4797
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.439121346543005e-08,
+      "loss": 0.1442,
+      "step": 4798
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.382742318610298e-08,
+      "loss": 0.0759,
+      "step": 4799
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3270219999790286e-08,
+      "loss": 0.119,
+      "step": 4800
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.271960415156926e-08,
+      "loss": 0.0866,
+      "step": 4801
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.217557588362451e-08,
+      "loss": 0.092,
+      "step": 4802
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.1638135435239626e-08,
+      "loss": 0.1204,
+      "step": 4803
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.1107283042798854e-08,
+      "loss": 0.0582,
+      "step": 4804
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.058301893979542e-08,
+      "loss": 0.1146,
+      "step": 4805
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.00653433568182e-08,
+      "loss": 0.0681,
+      "step": 4806
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.955425652156173e-08,
+      "loss": 0.1112,
+      "step": 4807
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.904975865882119e-08,
+      "loss": 0.0803,
+      "step": 4808
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8551849990494085e-08,
+      "loss": 0.078,
+      "step": 4809
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8060530735580227e-08,
+      "loss": 0.1196,
+      "step": 4810
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7575801110178424e-08,
+      "loss": 0.1209,
+      "step": 4811
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.709766132749313e-08,
+      "loss": 0.1488,
+      "step": 4812
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6626111597829454e-08,
+      "loss": 0.1068,
+      "step": 4813
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.616115212858982e-08,
+      "loss": 0.0651,
+      "step": 4814
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5702783124285636e-08,
+      "loss": 0.0502,
+      "step": 4815
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.525100478652064e-08,
+      "loss": 0.0642,
+      "step": 4816
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4805817314005877e-08,
+      "loss": 0.1003,
+      "step": 4817
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4367220902553046e-08,
+      "loss": 0.0968,
+      "step": 4818
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3935215745072838e-08,
+      "loss": 0.0583,
+      "step": 4819
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.350980203157659e-08,
+      "loss": 0.093,
+      "step": 4820
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3090979949177962e-08,
+      "loss": 0.0985,
+      "step": 4821
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2678749682092928e-08,
+      "loss": 0.0987,
+      "step": 4822
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.227311141163312e-08,
+      "loss": 0.0511,
+      "step": 4823
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1874065316214155e-08,
+      "loss": 0.1112,
+      "step": 4824
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1481611571353968e-08,
+      "loss": 0.093,
+      "step": 4825
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.109575034966781e-08,
+      "loss": 0.087,
+      "step": 4826
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0716481820871593e-08,
+      "loss": 0.1281,
+      "step": 4827
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0343806151781876e-08,
+      "loss": 0.0803,
+      "step": 4828
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.977723506315873e-09,
+      "loss": 0.0923,
+      "step": 4829
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.618234045493113e-09,
+      "loss": 0.1333,
+      "step": 4830
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.265337927425455e-09,
+      "loss": 0.0656,
+      "step": 4831
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.9190353073354e-09,
+      "loss": 0.0921,
+      "step": 4832
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.579326337537774e-09,
+      "loss": 0.0761,
+      "step": 4833
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.246211167449724e-09,
+      "loss": 0.1138,
+      "step": 4834
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.919689943587382e-09,
+      "loss": 0.0703,
+      "step": 4835
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.5997628095692e-09,
+      "loss": 0.0949,
+      "step": 4836
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.286429906107617e-09,
+      "loss": 0.076,
+      "step": 4837
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.9796913710223895e-09,
+      "loss": 0.0797,
+      "step": 4838
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.679547339228931e-09,
+      "loss": 0.0963,
+      "step": 4839
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.385997942738308e-09,
+      "loss": 0.103,
+      "step": 4840
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.09904331066724e-09,
+      "loss": 0.0813,
+      "step": 4841
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.818683569229765e-09,
+      "loss": 0.0645,
+      "step": 4842
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.544918841738911e-09,
+      "loss": 0.0782,
+      "step": 4843
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.277749248605024e-09,
+      "loss": 0.1171,
+      "step": 4844
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.017174907339106e-09,
+      "loss": 0.1055,
+      "step": 4845
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.763195932554476e-09,
+      "loss": 0.0481,
+      "step": 4846
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.515812435960109e-09,
+      "loss": 0.0944,
+      "step": 4847
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.275024526363969e-09,
+      "loss": 0.0973,
+      "step": 4848
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.040832309673004e-09,
+      "loss": 0.098,
+      "step": 4849
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.813235888894817e-09,
+      "loss": 0.0683,
+      "step": 4850
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.5922353641359986e-09,
+      "loss": 0.0639,
+      "step": 4851
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.3778308325987937e-09,
+      "loss": 0.1274,
+      "step": 4852
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.1700223885877676e-09,
+      "loss": 0.0844,
+      "step": 4853
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.968810123504806e-09,
+      "loss": 0.0792,
+      "step": 4854
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7741941258524472e-09,
+      "loss": 0.0985,
+      "step": 4855
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5861744812272214e-09,
+      "loss": 0.1019,
+      "step": 4856
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.4047512723296418e-09,
+      "loss": 0.1132,
+      "step": 4857
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.2299245789542123e-09,
+      "loss": 0.1036,
+      "step": 4858
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.0616944779994205e-09,
+      "loss": 0.0791,
+      "step": 4859
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.900061043457746e-09,
+      "loss": 0.0954,
+      "step": 4860
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7450243464223191e-09,
+      "loss": 0.1144,
+      "step": 4861
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5965844550819287e-09,
+      "loss": 0.071,
+      "step": 4862
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.4547414347293452e-09,
+      "loss": 0.0621,
+      "step": 4863
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3194953477496663e-09,
+      "loss": 0.088,
+      "step": 4864
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.190846253631972e-09,
+      "loss": 0.0789,
+      "step": 4865
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.068794208959334e-09,
+      "loss": 0.0721,
+      "step": 4866
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.533392674154762e-10,
+      "loss": 0.1139,
+      "step": 4867
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.444814797814449e-10,
+      "loss": 0.089,
+      "step": 4868
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.422208939389385e-10,
+      "loss": 0.0909,
+      "step": 4869
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.465575548619818e-10,
+      "loss": 0.1084,
+      "step": 4870
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.574915046319129e-10,
+      "loss": 0.0869,
+      "step": 4871
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.750227824190656e-10,
+      "loss": 0.0643,
+      "step": 4872
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.991514244994221e-10,
+      "loss": 0.1033,
+      "step": 4873
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.2987746424295586e-10,
+      "loss": 0.1343,
+      "step": 4874
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.672009321186275e-10,
+      "loss": 0.0847,
+      "step": 4875
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.111218556927197e-10,
+      "loss": 0.0946,
+      "step": 4876
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6164025963549823e-10,
+      "loss": 0.0911,
+      "step": 4877
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1875616570622415e-10,
+      "loss": 0.0969,
+      "step": 4878
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.246959276814181e-11,
+      "loss": 0.1006,
+      "step": 4879
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2780556783482705e-11,
+      "loss": 0.0988,
+      "step": 4880
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.968907080846961e-11,
+      "loss": 0.065,
+      "step": 4881
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.319514499997787e-11,
+      "loss": 0.0747,
+      "step": 4882
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.2987866122047296e-12,
+      "loss": 0.0803,
+      "step": 4883
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0,
+      "loss": 0.1038,
+      "step": 4884
+    },
+    {
+      "epoch": 2.0,
+      "step": 4884,
+      "total_flos": 321794859163648.0,
+      "train_loss": 0.21580008390061345,
+      "train_runtime": 35404.2594,
+      "train_samples_per_second": 6.621,
+      "train_steps_per_second": 0.138
+    }
+  ],
+  "max_steps": 4884,
+  "num_train_epochs": 2,
+  "total_flos": 321794859163648.0,
+  "trial_name": null,
+  "trial_params": null
+}